EGU Leonardo Conference Series on the Hydrological Cycle

6th Leonardo Conference 2014

Book of Abstracts

13–14 November 2014

Prague, Czech Republic

Organized by:

European Geophysical Union

UNESCO FRIEND-Water Program

T. G. Masaryk Water Research Institute, p.r.i., Prague

Czech University of Life Sciences, Prague

University of Natural Resources and Life Sciences, Vienna (BOKU)

Vienna University of Technology

Central Institute of Meteorology and Geodynamics, Vienna

Cite as:

Vizina, A. & Laaha, G. (Eds.) (2014): HYPER Droughts: Hydrological Pre-cipitation Evaporation Runoff Droughts - Book of Abstracts of the 6th EGULeonardo Conference, 13-14 Nov 2014, Prague, Czech Republic, 139 p.

Foreword

The Leonardo Conferences on Earth’s Hydrological Cycle is a well-establishedconference series of the European Geosciences Union (EGU) that focuses on top-ical questions related to water and its interactions with environment and society,with particular attention for water resources management, flood risk mitigationand environmental protection. After five conferences around the hydrologicaldimensions of remote sensing (Frascati, 2009), catchments (Luxembourg, 2010),floods (Bratislava, 2011), society and policy (Torino, 2012), and uncertainty(Kos, 2013), the sixth conference held from 13-14 November 2014 in Prague(Czech Republic) was dedicated again to a hydrological extreme: hydrologicaldrought within the various, connected components of the water cycle.

Under the title HYPER Droughts (Hydrological Precipitation – Evaporation– Runoff Droughts), experts from a broad number of disciplines, includingsurface- and groundwater hydrology, climatology and meteorology, (hydro-)ecologyand water quality, geography and history, wood science, water monitoring andmanagement, and environmental statistics, were brought together to comparethe current status of drought research in these disciplines, and to assess the im-plications of drought for water resources management.

The conference was organised around the following topics:

1. Drought governing processes including climate and catchment drivers

2. Indices and tools for monitoring and prediction

3. Regional modelling of low flows and droughts

4. Reconstruction of historic and prehistoric droughts (from modelling, doc-umentary sources and other proxies, such as tree rings)

5. Hydrological change (climate change and human impacts)

6. Drought forecasting (meteorological, hydrological and agricultural perspec-tives)

7. Drought impacts (ecological and economic)

8. Implications for water management

This volume bundles the abstracts of 2 Leonardo lectures, 4 keynote lectures,30 oral presentations and 39 poster presentations from over 31 countries whichreflect the international dimension of the low flow and drought topic. Our thanksgo to the organizing committee of the Czech University of Life Sciences, Prague,

1

T. G. Masaryk Water Research Institute, p.r.i., Prague and the European LowFlow and Drought group of the UNESCO FRIEND-Water program for makingthis event possible.

We hope that the conference will stimulate international cooperation andinitiatives, crossing national boundaries and disciplines and, thereby, will con-tribute to a better assessment of our water resources under current, past andfuture environmental conditions.

Gregor LaahaInstitute of Applied Statistics and Computing

University of Natural Resources and Life Sciences, Vienna (BOKU)

2

The Venue of the Conference

The conference will take place at the Faculty of Environmental Sciences, CzechUniversity of Life Sciences Prague (FES). The FES is located in the CULSPrague Campus. The campus is about 30 minutes from the city centre (us-ing the Prague Public transportation). Further information about venue andaccommodation are available from: http://www.eguleonardo2014.com

Organization

– The official language of the conference is English.

– Sessions will be designed around the conference topics. Both oral andposter presentations are welcome.

– A summary report of the main findings and recommendations from eachsession will be given and discussed in a closing plenary session.

– Selected presentations will be considered for publication in an EGU journal(special issue) following the conference.

– Participation/registration fee will be waived on request to speakers fromdeveloping countries and students. Participants are expected to financetheir own travel and accommodation expenses.

– Option for project meetings before the conference (Wednesday), handledon a first-come first-served basis.

Registration fees

Normal 210 EURStudent (full) 110 EURStudent (without dinner) 70 EURDeveloping countries (without dinner) 70 EUR

For speakers from developing countries, a waiver (0 EUR) may be offered onrequest.

The (full) registration fees include:

– Admission to all congress sessions

– All congress materials

3

– Daily coffee/tea breaks

– Lunches on November 13th, 14th

– Boat trip & Gala dinner on November 13th

Presentations

Oral Presentations

The speakers will have 15 min to present their paper, followed by 5 min ofquestions. The presentations will be made in English. A detailed programincluding titles of oral presentations is provided below.

Posters

Posters will be shown in a dedicated room on Thursday 13 and Wednesday 14November.

Part of the presentations will take place in parallel sessions by topic, in tworooms and will be organized as follows:

4

Wednesday 12 November

19.00 Icebreaker

Thursday 13 November

Time Room A Room B

07.30–08.30 Registration

08.30–09.00 Opening session

09.00–09.30 Keynote: Anne van Loon

09.30–10.30 Plenary 1: Drought governing processes includingclimate and catchment drivers

10.30–11.00 Coffee break

11.00–11.30 Keynote: Miroslav Trnka

11.40–12.40 Session 2A: Indices and toolsfor monitoring and prediction

Session 2B: Drought govern-ing processes including cli-mate and catchment drivers

12.40–13.45 Lunch break

13.45–14.45 Session 3A: Reconstructionof historic and prehistoricdroughts

Session 3B: Indices and toolsfor monitoring and prediction

14.45–15.15 Coffee break

15.15–15.45 Discussion 1

15.45–17.30 Poster session

18.30– Boat trip & Gala dinner

5

Friday 14 November

Time Room A Room B

08.30–10.00 Plenary 4: Hydrological change (climate changeand human impacts)

08.30–09.00 Keynote: Christel Prudhomme

09.00–10.00 Oral presentations

10.10–11.10 Session 5A: Hydrologicalchange

Session 5B: Regional mod-elling of low flows anddroughts

11.10–11.30 Coffee break

11.30–12.30 Session 6A: Drought forecast-ing

Session 6B: Drought impacts

12.30–13.30 Lunch break

13.30–14.45 Leonardo Lectures 201413.35 Lena M. Tallaksen (Norway)14.15 Klaus Haslinger (Austria)

14.45–15.15 Coffee break

15.15–16.25 Plenary 7: Implications for water management)15.15 Keynote: Mark Svoboda15.45 Oral presentations

16.25–16.55 Discussion 2

16.55–17.00 Closing session

6

International Scientific Committee

Gregor Laaha, BOKU University Vienna, Austria (Chairperson)

Mary-Jane Adler, National Institute of Hydrology and Water Management,Romania

Gunter Bloschl, Vienna University of Technology, Austria

Miriam Fendekova, Comenius University Bratislava, Slovakia

Jamie Hannaford, Centre for Ecology and Hydrology, Wallingford, UK

Claire Lang-Delus, University of Lorraine, France

Henny van Lanen, Wageningen University, the Netherlands

Anne van Loon, Wageningen University, the Netherlands

Alberto Montanari, University of Bologna, Italy

Pavel Pech, Czech University of Life Sciences, Prague, Czech Republic

Wolfgang Schoner, Central Institute of Meteorology and Geodynamics, Vi-enna, Austria

Sonia Seneviratne, ETH Zurich, Switzerland

Justin Sheffield, Princeton University, USA

Petr Sklenicka, Czech University of Life Sciences, Prague, Czech Republic

Karsten Schulz, BOKU University Vienna, Austria

Kerstin Stahl, University of Freiburg, Germany

Lena Tallaksen, University Oslo, Norway

Jean-Philippe Vidal, Irstea (formerly Cemagref), France

Local Organizing Committee

Adam Vizina, T. G. Masaryk Water Research Institute, Prague (Chairperson)

Martin Hanel, T. G. Masaryk Water Research Institute, Prague

Jirı Pavlasek, Czech University of Life Sciences, Prague

Petr Maca, Czech University of Life Sciences, Prague

Jan Danhelka, Czech Hydrometeorological Institute

7

Maps

8

HYPER Droughts Conference Program

Wednesday, November 12th, 2014

19.00 Icebreaker

Thursday, November 13th, 2014

07.30–08.30 Registration

08.30–09.00 Opening sessionJan Danhelka, Director of Department of hydrology, CzechHydrometeorological InstituteMark Rieder, Director of T. G. Masaryk Water ResearchInstitute, p.r.i.Gregor Laaha, EGU/HS

09.00–09.30 KEYNOTE: How to characterize hydrological droughtAnne F. van Loon

09.30–10.30 Plenary P1: Drought governing processes includingclimate and catchment driversChairperson: Gregor Laaha

09.30–09.50 Climate drivers of regional drought in EuropeJames Howard Stagge, Daniel G. Kingston, LenaMerete Tallaksen, David M. Hannah

09.50–10.10 Sensitivity of stream flow droughts and water scarcityevents to ENSO driven inter-annual climate variability atthe global scaleTed Veldkamp, Stephanie Eisner, Yoshihide Wada,Jeroen Aerts, Philip Ward

10.10–10.30 Recharge scenarios to identify controls of catchments’ sen-sitivity to droughtMichael Stoelzle, Kerstin Stahl, Andreas Morhard,Markus Weiler

10.30–11.00 Coffee break

9

11.00–11.30 KEYNOTE: Integrated drought monitoring sys-tems – a tool for drought research and practicalapplicationsMiroslav Trnka

11.40–12.40 Session 2A: Indices and tools for monitoring andpredictionChairperson: Kerstin Stahl

11.40–12.00 Characterisation of groundwater droughtJohn Bloomfield, Ben P. Marchant

12.00–12.20 An alternative definition and characterisation of droughtterminationSimon Parry, Christel Prudhomme, Robert Wilby, PaulWood

12.20–12.40 Linking drought indicators with impacts: Insights from acase study in Germany and the UKSophie Bachmair, Kerstin Stahl, Jamie Hannaford,Lucy J. Barker, Irene Kohn, Cecilia Svensson, MalikoTanguy

11.40–12.40 Session 2B: Drought governing processes includingclimate and catchment driversChairperson: Martin Hanel

11.40–12.00 Influence of seasonal snowpack on summer low flowsMichal Jenicek, Jan Seibert, Massimiliano Zappa, MariaStaudinger, Tobias Jonas

12.00–12.20 Relationship between groundwater level and NDVI: sum-mer drought development analysed by insitu and MODIS-derived dataUrszula Somorowska

12.20–12.40 Differentiating between influencing factors land use and cli-mate to assess drought effects on groundwater recharge ina temperate contextBoud Verbeiren, Marijke Huysmans, Sven Vanderhae-gen, Frank Canters, Klaartje Verbeeck, Guy Engelen,Ingrid Jacquemin, Bernard Tychon, George Tsakiris,Harris Vangelis, Okke Batelaan

10

12.40–13.45 Lunch break

13.45–14.45 Session 3A: Reconstruction of historic and prehis-toric droughtsChairperson: Jean-Philippe Vidal

13.45–14.05 Reconstruction of meteorological droughts in France since1871 through the probabilistic downscaling of a global at-mospheric reanalysisLaurie Caillouet, Jean-Philippe Vidal, Eric Sauquet,Muriel Haond

14.05–14.25 Tree rings as a tool for reconstructing historic droughts innortheastern AustriaSandra Karanitsch-Ackerl, Michael Grabner, FranzHolawe, Gregor Laaha

14.25–14.45 Developing a multidisciplinary inventory of UK droughtsfrom the late 19th centuryJamie Hannaford, John Bloomfield, Ian Holman, Bet-tina Lange, Ralph Ledbetter, Mark McCarthy, TonyMcEnery, Rebecca Pearce, Steven Wade

13.45–14.45 Session 3B: Indices and tools for monitoring andpredictionChairperson: Miriam Fendekova

13.45–14.05 The analysis of storage discharge relationship for the iden-tification of low flows on selected watersheds in the CzechRepublicPetr Maca, Adam Vizina, Stanislav Horacek, MartinHanel, Ladislav Kasparek, Pavel Pech

14.05–14.25 Current and future drought vulnerability assessment in theJucar River basin SpainDavid Haro-Monteagudo, Joaquın Andreu, Abel Solera,Javier Paredes-Arquiola

14.25–14.45 —

14.45–15.15 Coffee break

11

15.15–15.45 Discussion of first block of sessions: Processes, In-dices and tools for monitoring and prediction, Re-constructionPodium: Kerstin Stahl, Anne van Loon, MiroslavTrnka, Jamie Hannaford, Jean-Philippe Vidal

15.45–17:30 Poster session

18.30–23.30 Boat trip & Gala dinnerMeeting at 18.30 at entrance to university campus (bus),return by public transport

Friday, November 14th, 2014

08.30–09.00 KEYNOTE: Droughts in the 21st century: a globalperspectiveChristel Prudhomme

09.00–10.00 Plenary 4: Hydrological changeChairperson: Jamie Hannaford

09.00–09.20 Future hydrological drought on a global and continentalscale: overview and outlookHenny van Lanen, Niko Wanders, Marjolein H. J. vanHuijgevoort, Yoshihide Wada, Marcel A. A. Alderli-este

09.20–09.40 Hierarchy of climate and hydrological uncertainties in lowflow projectionsJean-Philippe Vidal, Eric Sauquet, Claire Magand,Agnes Ducharne, Benoıt Hingray

09.40–10.00 Are runoff changes in Europe attributable to anthropogenicclimate changeLukas Gudmundsson

10.10–11.10 Session 5A: Hydrological changeChairperson: Lukas Gudmundsson

10.10–10.30 Impacts of Anthropogenic Emissions on Hydrologic CycleVenkatachalam Ramaswamy

12

10.30–10.50 Spatiotemporal drought analysis under varying climaticconditionsMarzena Osuch, Joanna Doroszkiewicz, Hadush KidaneMeresa, Renata Romanowicz

10.50–11.10 —

10.10–11.10 Session 5B: Regional modelling of low flows anddroughtsChairperson: Simon Parry

10.10–10.30 Statistical modeling of low flow conditions based on climaticindicatorsAnne Fangmann, Uwe Haberlandt

10.30–10.50 Spatial and temporal modelling of UK groundwaterdroughts using empirical modelsBen Marchant, John Bloomfield, Jon MacKay

10.50–11.10 Spatial and temporal variability of droughts in SwitzerlandGaetano di Franca, Peter Molnar, Paolo Burlando,Antonio Cancelliere

11.10–11.30 Coffee break

11.30–12.30 Session 6A: Drought forecastingChairperson: Petr Maca

11.30–11.50 The PREMHYCE project: a comparative evaluation of hy-drological models for low flow simulation and forecastingon French catchmentsPierre Nicolle, Raji Pushpalatha, Charles Perrin, Di-dier Francois, Dominique Thiery, Thibault Mathevet,Matthieu Le Lay, Francois Besson, Jean-Michel Soubey-roux, Christian Viel, Fabienne Regimbeau, VazkenAndreassian, Pascal Maugis, Benedicte Augeard, Em-manuel Morice

11.50–12.10 Trade-offs between dynamical and statistical forecasting ofthe European 2003 DroughtStephan Thober, Luis Samaniego, Rohini Kumar

13

12.10–12.30 Investigating the potential contribution of teleconnectionindices to forecast short term drought class transitions inSicily, ItalyBrunella Bonaccorso, Antonino Cancelliere, GiuseppeRossi

11.30–12.30 Session 6B: Drought impactsChairperson: James Howard Stagge

11.30–11.50 An impact perspective on pan-European drought sensitivityKerstin Stahl, Irene Kohn, Lucia de Stefano, Lena M.Tallaksen, Francisco M. C. Castro Rego, Dionysis Assi-macopoulos, Antonio Massarutto, Sonia I. Seneviratne,Henny H. A. J. van Lanen

11.50–12.10 BIODROUGHT project - developing effective tools for ret-rospective bioindication of dry episodes in stream historyand selection of watersheds with drought risk using aquaticmacroinvertebratesPetr Paril

12.10–12.30 Assessing future droughts economic losses with generic hy-droeconomic modellingNoemie Neverre, Patrice Dumas

12.30–13.30 Lunch break

13.30–14.45 Leonardo Lectures 2014Chairperson: Gregor Laaha

13.35–14.15 Lena M. Tallaksen (Norway)Title: Low flow and drought in Europe - lessons from thepast

14.15–14.45 Klaus Haslinger (Austria)Title: Droughts and low flows in present and future climate– pacing the path between climatology and hydrology

14:45–15.15 Coffee break

15.15–16.25 Plenary P7: Implications for water managementChairperson: Henny van Lanen

14

15.15–15.45 KEYNOTE: National Drought Mitigation CenterDrought Services: Collaborations Towards a GlobalDrought Early Warning and Information SystemMark Svoboda

15.45–16.05 Multisectorial Partnerships for drought management andmitigation: The experience of the Jucar River Basin SpainJoaquın Andreu, David Haro, Abel Solera, JavierParedes-Arquiola

16.05–16.25 Towards a comprehensive drought information system forEuropeJurgen Vogt

16.25–16.55 Discussion of second block of sessions: Modelling,impacts and implications for water managementPodium: Henny van Lanen, Lena M. Tallaksen, MarkSvoboda, Christel Prudhomme, Jurgen Vogt

16.55–17.00 Closing session

15

HYPER Droughts list of posters

Thursday, November 13th, 2014

15.45-17:30 Room RK

Poster number

1 Drought Duration Curves: A method to quantify continental differences in hydro-logical droughtsErik Tijdeman, Kerstin Stahl, Sophie Bachmair, Jamie Han-naford

2 The influence of catchment parameters on drought occurrence in the Tatra Mts,.SlovakiaMiriam Fendekova, Marian Fendek

3 An assessment of actual evapotranspiration and soil water deficit in agriculturalregions in EuropeBlaz Kurnik

4 Response of simple stream hydro-chemical indicators to rainfall-runoff events inforest disturbed basins in upper Vydra, Central Sumava MountainsYe Su, Jakub Langhammer, Jana Kaiglova, Jerker Jarsjo

5 Identifying a global drought structure using the copula methodologyFranziska Gaupp

6 Investigation of Drought Occurrence and Propagation between Meteorological,Agricultural, and Hydrological DroughtJi Yae Shin, Jiyoung Yoo, Minsung Kwon, Tae-Woong Kim

7 IA drought index accounting for snowMaria Staudinger, Jan Seibert, Kerstin Stahl

8 Identification of Consecutive Dry Days (CDD) using observational and model datafrom COSMO CLMLivia Labudova, Lukas Schefczyk, Milan Trizna

9 Spatio-temporal characteristics of wet and drought extremes over territory of Rus-siaDilyara Utkuzova

10 Assessing the drought risk based on the joint probability of precipitation and soilmoisture in Rhineland-Palatinate, southwest GermanyZhiyong Liu, Lucas Menzel

11 Stochastic modeling of spatio-temporal characteristics of drought in southern IranArash Malekian

16

12 Influence of the Water Channel Network Facility in the Kiso River SystemYoshinobu Sato

13 IA drought index accounting for snowJuraj Parajka, Jan Seibert, Kerstin Stahl

14 Tree rings as a tool for reconstructing historic droughts in northeastern AustriaSandra Karanitsch-Ackerl, Michael Grabner, Franz Holawe,Gregor Laaha

15 Developing a multi-disciplinary inventory of UK droughts from the late 19th cen-turyJamie Hannaford, John Bloomfield Ian Holman, BettinaLange, Ralph Ledbetter, Mark McCarthy, Tony McEnery,Rebecca Pearce, Steven Wade

16 Drought Signatures in the tree-ring records of European RussiaVladimir Matskovsky, Olga Solomina, Sergey Matveev,Alexander Chernokulsky

17 A comparative study of dryness indices over EuropeConcepcion Rodriguez-Puebla

18 Analysis of uncertainty in climate change projections of drought indices for theCzech RepublicMartin Hanel, Adam Vizina, Adri Buishand

19 Scaling of low flows in CzechiaOndrej Ledvinka

20 A Statistical Analysis of the temporal-spatial changing characteristics of extremeprecipitation in Yunnan Province from 1958-2013Xiao Jing Yang, Z. X. Xu, D. P. Zuo, L. Liu

21 Conceptual Framework for Satellite Data Based Drought Outlook in KoreaTae-Woong Kim, Ji Yaes Shin, Joo-Heon Lee, Hyun-Han Kwon

22 Current and future agricultural droughts in the southeastern Mediterranean regionLucas Menzel, Tobias Tornros

23 Towards pan–European drought risk maps: comparing an impact-based approachwith a factor-based vulnerability assessmentVeit Blauhut, Julia Urquijo, Itziar Gonzalez Tanago, KerstinStahl, Lucia De Stefano, Mario Ballesteros

24 Different approaches to drought capture – how do aquatic invertebrates indicatedry episode in Central European streams?Marek Polasek, Petr Paril, Svetlana Zahradkova, Vıt Sy-rovatka, Lenka Sikulova, Michal Straka, Denisa Nemejcova

17

25 Atmospheric drought in Belgium - Statistical analysis of precipitation deficitOrhan Dengiz, Marketa Mihalikova

26 Testing of Linear Trend Appearance in Precipitation and TemperatureRoman Kozın

27 A project of the hydrological drought monitoring in the Czech RepublicRadek Vlnas, Adam Beran

28 Drought propagation by using weather generatorsAdam Vizina, Martin Hanel, Ladislav Kasparek, Petr Maca,Pavel Pech

29 Drought propagation analysis using different threshold level methodsHeudorfer, Kerstin Stahl

30 Atmospheric drought in Belgium - Statistical analysis of precipitation deficitSepideh Zamani, Martin Hanel, Pavel Pech

31 The European Drought Impact report Inventory (EDII): major drought eventsIrene Kohn, J.H. Stagge, V. Blauhut, L. M. Tallaksen, S.Bachmair, K. Stahl

32 Investigating seasonal variables for droughts in cold climates: a comparative studyin Austria and NorwayStefan Willem Ploum, Anne Frederike van Loon

33 Application of trivariate Gumbel–Hougaard copula for short term prognosis ofextreme low-flow eventsWojciech Jakubowski

34 Optimization of the Standardized Precipitation Index (SPI) for operationaldrought monitoringWillem Maetens, Brecht Ingels, Willem Defloor, KrisCauwenberghs

35 A comprehensive and effective representation of the drought situation: implemen-tation through the web portal www.waterinfo.beWillem Maetens, Pieter Cabus, Inge De Jongh, Willem De-floor, Kris Cauwenberghs

36 SWAT-modelling of drought impact on two catchments in FlandersWillem Maetens, Brecht Ingels, Willem Defloor, KrisCauwenberghs

37 Analysis of two climatological data sources to apply to description of the precipi-tation behavior in an arid area: Case of Region of Tarapaca, Northern ChileDamian Ferney Cordoba Melgarejo, Concepción RodrıguezPuebla, Jazna Rodrıguez Segovia

18

38 Evaluation of the socio-economic impacts of drought events: evidence from someareas in EuropeDario Musolino

39 An application of remote sensing to support to water resources managementFrancesco Vuolo

40 Assessing and monitoring the 2005 and 2012 drought events in Iberia using vege-tation and drought indicesCelia Gouveia, Ana Russo Ricardo Machado Trigo

41 Comparison of SOM and non-hierarchic clustering techniques by pooling of lowflows in SlovakiaSilvia Kohnova, Miroslav Sabo, Andrea Stevkova, KatarınaJeneiova, Kamila Hlavcova, Jan Szolgay

42 The inadequacy of the standardized precipitation index as a groundwater droughtindicatorJude Lubega Musuuza, Rohini Kumar, Anne F. Van Loon,Roland Barthel, Jurriaan ten Broek, Luis E. Samaniego,Adriaan J. Teuling, Sabine Attinger

43 Flow-duration-frequency and duration-flow-frequency approaches in low-flow anal-ysis, advantages and constraintsEwa Bogdanowicz, Witold Gustaw Strupczewski, Sisay De-bele, Krzysztof Kochanek

19

Contents

Anne Van Loon : How to characterise hydrological drought? . . . . . . . . 27James Howard Stagge, Daniel G. Kingston, Lena Merete Tallaksen, David

M. Hannah : Climate drivers of regional drought in Europe . . . . 29Ted Veldkamp, Stephanie Eisner, Yoshihide Wada, Jeroen Aerts Philip Ward

: Sensitivity of stream flow droughts and water scarcity events to

ENSO driven inter-annual climate variability at the global scale . . 31Michael Stoelzle, Kerstin Stahl, Andreas Morhard, Markus Weiler : Recharge

scenarios to identify controls of catchments’ sensitivity to drought . 33Miroslav Trnka, : Integrated drought monitoring systems – a tool for drought

research and practical applications . . . . . . . . . . . . . . . . 34John P. Bloomfield, Ben P. Marchant : Characterisation of groundwater drought 35Simon Parry, Christel Prudhomme, Robert Wilby, Paul Wood : An alterna-

tive definition and characterisation of drought termination . . . . . 37Sophie Bachmair, Kerstin Stahl, Jamie Hannaford, Lucy J. Barker, Irene

Kohn, Cecilia Svensson, Maliko Tanguy : Linking drought indicators

with impacts: Insights from a case study in Germany and the UK . 39Michal Jenicek, Jan Seibert, Massimiliano Zappa, Maria Staudinger, Tobias

Jonas : Influence of seasonal snowpack on summer low flows . . . . 41Urszula Somorowska : Relationship between groundwater level and NDVI:

summer drought development analysed by in-situ and MODIS-derived

data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Boud Verbeiren, Marijke Huysmans, Sven Vanderhaegen, Frank Canters,

Klaartje Verbeeck, Guy Engelen, Ingrid Jacquemin, Bernard Ty-

chon, George Tsakiris, Harris Vangelis, Okke Batelaan : Differen-

tiating between influencing factors land use and climate to assess

drought effects on groundwater recharge in a temperate context . . 45Laurie Caillouet, Jean-Philippe Vidal, Eric Sauquet, Muriel Haond : Recon-

struction of meteorological droughts in France since 1871 through

the probabilistic downscaling of a global atmospheric reanalysis . . 47Sandra Karanitsch-Ackerl, Michael Grabner, Franz Holawe, Gregor Laaha :

Tree rings as a tool for reconstructing historic droughts in northeast-

ern Austria . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

20

Jamie Hannaford John Bloomfield Ian Holman Bettina Lange Ralph Ledbet-

ter Mark McCarthy Tony McEnery Rebecca Pearce Steven Wade :

Developing a multi-disciplinary inventory of UK droughts from the

late 19th century . . . . . . . . . . . . . . . . . . . . . . . . 51

Petr Maca, Adam Vizina, Stanislav Horacek, Martin Hanel, Ladislav Kas-

parek, Pavel Pech : The analysis of storage discharge relationship for

the identification of low flows on selected watersheds in the Czech

republic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

David Haro, Abel Solera, Javier Paredes-Arquiola, Joaquín Andreu : Current

and future drought vulnerability assessment in the Jucar River basin

(Spain) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Christel Prudhomme, : Droughts in the 21st century: a global perspective . 57

Henny A.J. van Lanen, Niko Wanders, Marjolein H.J. van Huijgevoort, Yoshi-

hide Wada, Marcel A. A. Alderlieste : Future hydrological drought

on a global and continental scale: overview and outlook . . . . . . 58

Jean-Philippe Vidal, Eric Sauquet, Claire Magand, Agnès Ducharne, Benoît

Hingray : Hierarchy of climate and hydrological uncertainties in low

flow projections . . . . . . . . . . . . . . . . . . . . . . . . . 60

Venkatachalam Ramaswamy : Impacts of Anthropogenic Emissions on Hy-

drologic Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Lukas Gudmundsson : Are runoff changes in Europe attributable to anthro-

pogenic climate change? . . . . . . . . . . . . . . . . . . . . . 63

Marzena Osuch, Joanna Doroszkiewicz, Hadush Kidane Meresa, Renata Ro-

manowicz : Spatio-Temporal Drought Analysis Under Varying Cli-

matic Conditions . . . . . . . . . . . . . . . . . . . . . . . . 64

Anne Fangmann, Uwe Haberlandt : Statistical Modeling of Low Flow Con-

ditions based on Climatic Indicators . . . . . . . . . . . . . . . 65

Ben Marchant, John Bloomfield, Jon MacKay : Spatial and temporal mod-

elling of UK groundwater droughts using empirical models . . . . . 66

Gaetano Di Franca, Peter Molnar, Paolo Burlando, Antonio Cancelliere :

Spatial and temporal variability of droughts in Switzerland . . . . 68

21

Pierre Nicolle, Raji Pushpalatha, Charles Perrin, Didier François, Dominique

Thiéry, Thibault Mathevet, Matthieu Le Lay, François Besson, Jean-

Michel Soubeyroux, Christian Viel, Fabienne Regimbeau, Vazken

Andréassian, Pascal Maugis, Bénédicte Augeard, Emmanuel Morice :

The PREMHYCE project: a comparative evaluation of hydrological

models for low-flow simulation and forecasting on French catchments 70

Stephan Thober, Luis Samaniego, Rohini Kumar : Trade-Offs between dy-

namical and statistical forecasting of the European 2003 Drought . 72

Brunella Bonaccorso, Antonino Cancelliere, Giuseppe Rossi : Investigating

The Potential Contribution Of Teleconnection Indices To Forecast

Short Term Drought Class Transitions In Sicily (italy) . . . . . . 74

Kerstin Stahl, Irene Kohn, Lucia De Stefano, Lena M. Tallaksen, Francisco

M.C. Castro Rego, Dionysis Assimacopoulos, Antonio Massarutto,

Sonia I. Seneviratne, Henny H.A.J. van Lanen : An impact perspec-

tive on pan-European drought sensitivity . . . . . . . . . . . . . 75

Petr Paril, Světlana Zahrádková, Marek Polášek, Vít Syrovátka, Michal Straka,

Lenka Šikulová, Denisa Němejcová, Pavla Řezníčková : BIODROUGHT

project – developing effective tools for retrospective bioindication

of dry episodes in stream history and selection of watersheds with

drought risk using aquatic macroinvertebrates . . . . . . . . . . 76

Noémie Neverre, Patrice Dumas : Assessing future droughts economic losses

with generic hydroeconomic modelling . . . . . . . . . . . . . . 78

Lena M. Tallaksen : Low flow and drought in Europe - lessons from the past 80

Klaus Haslinger : Droughts and low flows in present and future climate –

pacing the path between climatology and hydrology . . . . . . . . 82

Mark D. Svoboda : National Drought Mitigation Center Drought Services:

Collaborations Towards a Global Drought Early Warning and Infor-

mation System . . . . . . . . . . . . . . . . . . . . . . . . . 83

David Haro, Abel Solera, Javier Paredes-Arquiola, Joaquín Andreu : Multi-

sectorial Partnerships for drought management and mitigation. The

experience of the Jucar River Basin (Spain) . . . . . . . . . . . . 84

22

Jürgen Vogt A. de Jager, D. Magni, L. Valentini, F. Micale, C. Cammalleri,

G. Sepulcre, C. Lavaysse, S. Russo, P. Barbosa, J. Spinoni : Towards

a comprehensive drought information system for Europe . . . . . . 86Erik Tijdeman, Kerstin Stahl, Sophie Bachmair, Jamie Hannaford : Drought

Duration Curves: A method to quantify continental differences in

hydrological droughts . . . . . . . . . . . . . . . . . . . . . . 88Miriam Fendekova, Marian Fendek : The influence of catchment parameters

on drought occurrence in the Tatra Mts., Slovakia . . . . . . . . . 90Blaz Kurnik : An assessment of actual evapotranspiration and soil water

deficit in agricultural regions in Europe . . . . . . . . . . . . . . 91Ye Su, Jakub Langhammer, Jana Kaiglova, Jerker Jarsjö : Response of simple

stream hydro-chemical indicators to rainfall-runoff events in forest

disturbed basins in upper Vydra, Central Sumava Mountains . . . 92Franziska Gaupp : Identifying a global drought structure using the copula

methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 94Ji Yae Shin, Jiyoung Yoo, Minsung Kwon, Tae-Woong Kim : Investigation of

Drought Occurrence and Propagation between Meteorological, Agri-

cultural, and Hydrological Drought . . . . . . . . . . . . . . . . 96Maria Staudinger, Jan Seibert, Kerstin Stahl : A drought index accounting

for snow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97Livia Labudova, Lukas Schefczyk, Milan Trizna : Identification of Consec-

utive Dry Days (CDD) using observational and model data from

COSMO CLM . . . . . . . . . . . . . . . . . . . . . . . . . . 98Dilyara Utkuzova : Spatio-temporal characteristics of wet and drought ex-

tremes over territory of Russia . . . . . . . . . . . . . . . . . . 99Zhiyong Liu, Lucas Menzel : Assessing the drought risk based on the joint

probability of precipitation and soil moisture in Rhineland-Palatinate,

southwest Germany . . . . . . . . . . . . . . . . . . . . . . . 100Arash Malekian : Stochastic modeling of spatio-temporal characteristics of

drought in southern Iran . . . . . . . . . . . . . . . . . . . . . 101Yoshinobu Sato : Influence of the Water Channel Network Facility in the

Kiso River System . . . . . . . . . . . . . . . . . . . . . . . . 103

23

Juraj Parajka, Alberto Viglione, Günter Blöschl : Rainfall-runoff projections

of low flows in Austria . . . . . . . . . . . . . . . . . . . . . . 105Vladimir Matskovsky, Olga Solomina, Sergey Matveev, Alexander Chernokul-

sky : Drought Signatures in the tree-ring records of European Russia 107Concepcion Rodriguez-Puebla : A comparative study of dryness indices over

Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109Martin Hanel, Adam Vizina, Adri Buishand : Analysis of uncertainty in

climate change projections of drought indices for the Czech Republic 110Ondrej Ledvinka : Scaling of low flows in Czechia . . . . . . . . . . . . . 111Xiao Jing Yang, Z. X. Xu, D. P. Zuo, L. Liu : A Statistical Analysis of the

temporal-spatial changing characteristics of extreme precipitation in

Yunnan Province from 1958-2013 . . . . . . . . . . . . . . . . . 113Tae-Woong Kim Ji Yaes Shin Joo-Heon Lee Hyun-Han Kwon : Conceptual

Framework for Satellite Data Based Drought Outlook in Korea . . 115Lucas Menzel Tobias Törnros : Current and future agricultural droughts in

the southeastern Mediterranean region . . . . . . . . . . . . . . 116Veit Blauhut Julia Urquijo Itziar Gonzalez Tanago Kerstin Stahl Lucia De

Stefano Mario Ballesteros : Towards pan–European drought risk

maps: comparing an impact-based approach with a factor-based vul-

nerability assessment . . . . . . . . . . . . . . . . . . . . . . 118Marek Polášek Petr Pařil Světlana Zahrádková Vít Syrovátka Lenka Šikulová

Michal Straka Denisa Němejcová : Different approaches to drought

capture – how do aquatic invertebrates indicate dry episode in Cen-

tral European streams? . . . . . . . . . . . . . . . . . . . . . 119Orhan Dengiz, Markéta Miháliková : Atmospheric drought in Belgium - Sta-

tistical analysis of precipitation deficit . . . . . . . . . . . . . . 121Roman Kožín : Testing of Linear Trend Appearance in Precipitation and

Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . 122Radek Vlnas, Adam Beran : A project of the hydrological drought monitoring

in the Czech Republic . . . . . . . . . . . . . . . . . . . . . . 124Adam Vizina, Martin Hanel, Ladislav Kašpárek, Petr Máca, Pavel Pech :

Drought propagation by using weather generators . . . . . . . . . 125

24

Benedikt Heudorfer, Kerstin Stahl : Drought propagation analysis using dif-

ferent threshold level methods . . . . . . . . . . . . . . . . . . 126

Sepideh Zamani : Atmospheric drought in Belgium - Statistical analysis of

precipitation deficit . . . . . . . . . . . . . . . . . . . . . . . 127

Irene Kohn, J.H. Stagge, V. Blauhut, L. M. Tallaksen, S. Bachmair, K. Stahl

: The European Drought Impact report Inventory (EDII): major

drought events . . . . . . . . . . . . . . . . . . . . . . . . . 128

Stefan Willem Ploum, Anne Frederike van Loon : Investigating seasonal vari-

ables for droughts in cold climates: a comparative study in Austria

and Norway . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

Wojciech Jakubowski : Application of trivariate Gumbel–Hougaard copula

for short term prognosis of extreme low-flow events . . . . . . . . 130

Willem Maetens, Brecht Ingels, Willem Defloor, Kris Cauwenberghs : Op-

timization of the Standardized Precipitation Index (SPI) for opera-

tional drought monitoring . . . . . . . . . . . . . . . . . . . . 131

Willem Maetens, Pieter Cabus, Inge De Jongh, Willem Defloor, Kris Cauwen-

berghs : A comprehensive and effective representation of the drought

situation: implementation through the web portal www.waterinfo.be 133

Willem Maetens, Brecht Ingels, Willem Defloor, Kris Cauwenberghs : SWAT-

modelling of drought impact on two catchments in Flanders . . . . 135

Damian Ferney Cordoba Melgarejo, Concepción Rodríguez Puebla, Jazna

Rodríguez Segovia : Analysis of two climatological data sources to

apply to description of the precipitation behavior in an arid area:

Case of Region of Tarapacá, Northern Chile . . . . . . . . . . . 137

Dario Musolino, : Evaluation of the socio-economic impacts of drought events:

evidence from some areas in Europe . . . . . . . . . . . . . . . 139

Francesco Vuolo : An application of remote sensing to support to water

resources management . . . . . . . . . . . . . . . . . . . . . . 140

Ewa Bogdanowicz Witold Gustaw Strupczewski Sisay Debele Krzysztof Kochanek

: Flow-duration-frequency and duration-flow-frequency approaches

in low-flow analysis, advantages and constraints . . . . . . . . . . 141

25

Kamila Hlavcova Silvia Kohnová Miroslav Sabo Andrea Stevková Jan Szolgay

: Comparison of SOM and non-hierarchic clustering techniques by

pooling of low flows in Slovakia . . . . . . . . . . . . . . . . . 142Jude Lubega Musuuza Rohini Kumar Anne F. Van Loon Roland Barthel

Jurriaan ten Broek Luis E. Samaniego Adriaan J. Teuling Sabine

Attinger : The inadequacy of the standardized precipitation index

as a groundwater drought indicator . . . . . . . . . . . . . . . . 143

26

How to characterise hydrological drought?

Anne Van Loon1

anne.vanloon@wur.nl

1 Wageningen University

General definitions of drought are “a sustained period of below-normal wateravailability” (Tallaksen and Van Lanen, 2004) and “a deficit of water comparedto normal conditions” (Sheffield and Wood, 2011). However, in the applicationof this definition, various interpretations are possible. Questions like: whatis sustained? what is normal?, need to be answered before droughts can bequantified. This applies to all types of drought including hydrological drought,i.e. drought in streamflow and hydrological stores (groundwater, lakes). Todraw conclusions based on quantitative information we need to understand theprocesses underlying hydrological drought development and use methods thatare suitable to represent these processes.

In this presentation, I will discuss the various climatic drivers of hydrologicaldrought and their interplay with catchment processes, leading to a number ofhydrological drought types. Storage of water in temporary stores, like snow andvegetation, or more permanent stores, like soil, groundwater, and glaciers, hasa major influence, both for average conditions (“normal”) and during extremeconditions (“drought”). The methods to define hydrological drought shouldideally take these processes into account, for example, by taking the seasonalcycle as “normal” condition. Also for statistical analysis of hydrological droughtcharacteristics, classification on the basis of a typology is useful so that droughtswith different governing factors are not confounded. In a study on global scalewe found, for example, that drought events of different type have deviatingdurations and deficit volumes.

To determine the “normal” situation and filter out extreme events, time se-ries of more than 30 years of discharge and groundwater level data are required.For attribution of drought in streamflow or groundwater also data of climaticdrivers are needed, at least precipitation and temperature. Unfortunately, suchlong records of hydrometeorological data are scarce. Especially, groundwaterdata are often lacking. Therefore, in most hydrological drought studies we must

27

rely on other data sources. Alternatives are the use of precipitation data di-rectly in drought indices like SPI (Standardised Precipitation Index) and theuse of simulated data of streamflow and groundater levels. Based on our recentstudies, I will demonstrate in which situations these alternatives can be appliedsuccessfully and what the pitfalls are. I show that the effects of delay and atten-uation by the stores (snow, groundwater) are often not well taken into account.This leads to discrepancies in terms of the temporal and spatial variation ofdrought.

Furthermore, I will discuss how other data sources can be useful in hydrolog-ical drought research. Modern data sources like satellite data (GRACE) mightbe a good alternative for groundwater level data, because the gravity signalrepresents the terrestrial storage of water. But direct use of the GRACE sig-nal deserves careful consideration in regions with a large influence of temporarystorage (e.g. snow). I also see potential in combining physical analysis with his-torical datasets and impact databases, like the US Drought Impact Reporter andthe European Drought Impact Report Inventory (EDII). Preliminary analyseshave indicated that the study of specific events can reveal underlying droughtprocesses.

References:Tallaksen, L. M., and Van Lanen, H. A. J., eds.: Hydrological drought: processes

and estimation methods for streamflow and groundwater, Developments in water sci-ence; 48, Elsevier Science B.V., Amsterdam, the Netherlands, 2004.

Sheffield, J., and Wood, E.: Drought; Past Problems and Future Scenarios, Earth-scan, London, UK, Washington DC, USA, 2011.

28

Climate drivers of regional drought in Europe

James Howard Stagge1, Daniel G. Kingston2, Lena Merete Tallaksen1, David M.Hannah3

j.h.stagge@geo.uio.no

1 University of Oslo, Department of Geosciences, Oslo, Norway2 Department of Geography, University of Otago, New Zealand

3 School of Geography, Earth and Environmental Sciences, University of Birmingham, UK

Better understanding is needed to characterize and quantify the effect of largescale climate drivers on extreme events such as droughts and floods. A previousstudy examined the major drivers of drought at the European scale, finding linkswith atmospheric circulation, wind speed, and sea surface temperature. Thesedrivers differ depending on whether drought severity is measured solely based onprecipitation or also including evapotranspiration. Analysis of drought at theEuropean scale has identified important climatic drivers, but considering theentire continent tends to smooth extremes and may mask important regionaldifferences. In particular, along the North-South gradient, the climate of themiddle and north of Europe is influenced year-round by the mid-latitude West-erlies, whereas the Mediterranean lies in a transitional zone, influenced by theSubtropical High Pressure Belt during the summer and mid-latitude Westerliesduring the winter. Along the West-East gradient, climatic influences transitionfrom the Atlantic to more Continental factors. This study expands on previouscontinental analyses by comparing drought patterns for homogenous sub-regionswithin Europe with hemispheric atmospheric circulation patterns. Focus nowincludes both high and low extremes (wet and dry); however, the following pre-sentation will focus only on droughts. Drought events are identified using twoindices, the 6-month cumulative Standardised Precipitation and StandardisedPrecipitation-Evapotranspiration Indices (SPI-6 and SPEI-6, respectively), bothcalculated using gridded Watch Forcing data (1958-2012), which combines theWatch Forcing and Watch Forcing ERA-Interim data sets. Regional climatolog-ical drought series (SPI, SPEI) are calculated for homogenous regions, definedby PCA and cluster analysis on SPI and SPEI simultaneously. Links betweenthe regional occurrence and severity of drought is analysed using two distinct ap-proaches. First, regional drought indices are compared with gridded geopotential

29

height fields and sea surface temperature, identifying regions of high correlationand creating atmospheric composites for the most extreme events. This providesa spatially detailed analysis. Second, the time series of regional climatologicaldrought is compared with potentially relevant circulation indices, such as theNAO, AO, and EA/WR, using wavelet analysis. This provides a detailed tem-poral analysis of common variation patterns. This study provides new insightinto how large-scale climatic variations affect the development and severity ofdrought in Europe, comparing the relevant drivers of drought across differentEuropean sub-regions and with previous findings for Europe as a whole.

30

Sensitivity of stream flow droughts and water scarcityevents to ENSO driven inter-annual climate variability at

the global scale

Ted Veldkamp1, Stephanie Eisner2, Yoshihide Wada3, Jeroen Aerts1, PhilipWard1

ted.veldkamp@vu.nl

1 Institute for Environmental Studies, VU University Amsterdam2 Center for Environmental Systems Research, University of Kassel

3 Department of Physical Geography, Utrecht University

Governments and institutions managing water resources have to adapt con-stantly to regional drought, water shortage and water stress conditions, beingcaused by climate change, socio-economic developments and/or climate variabil-ity. Taking into account the impact of climate variability is important as insome regions it may outweigh long-term climate change or socio-economic de-velopments, especially on a time scale of a few years up to a few decades. Asgovernments and water management institutions apply planning horizons up toa decade with respect to management of adaptation strategies, inter-annual cli-mate variability is especially relevant. A number of studies have estimated theimpacts of climate variability on stream flow droughts on a local, continental orglobal scale. Others have focused on the role of long term climate change andsocio-economic trends on blue water availability, and blue water scarcity. How-ever, a global assessment of the influence of inter-annual climate variability onstream flow droughts, blue water availability, and blue water scarcity togetherhas not yet been carried out, despite its importance for adaptation planning.

To address this issue, we assessed the influence of ENSO-driven climate vari-ability on stream flow droughts, blue water availability, and blue water scarcityevents at the global scale. Within this contribution we focused on El NinoSouthern Oscillation’s (ENSO) impact as ENSO is the most dominant sourceof inter-annual climate variability, impacting climate and society. We carriedout this assessment through the following steps: (1) used daily discharge andrun-off time-series (0.5o x 0.5o) of three WATCH-EI forced global hydrologi-cal models (Water-GAP, PCR-GLOBWB, and STREAM); (2) in combination

31

with time-series of population counts and monthly water demands we calculatedmonthly and yearly stream flow drought, water availability, water shortage andwater stress per Food Producing Unit (FPU) for the period 1960-2010; and(3) identified statistical relationships between stream flow droughts, blue wateravailability, blue water scarcity events and indices of ENSO in order to determineits sensitivity.

32

Recharge scenarios to identify controls of catchments’sensitivity to drought

Michael Stoelzle1, Kerstin Stahl1, Andreas Morhard2, Markus Weiler1

michael.stoelzle@hydrology.uni-freiburg.de

1 Chair of Hydrology University of Freiburg2 GIT HydroS Consult GmbH Freiburg, Germany

Streamflow drought depends to a large degree on groundwater recharge and stor-age depletion. During prolonged dry spells inherent catchment characteristicssuch as physiography and geology may be significant descriptors of streamflowresponse and baseflow-relevant subsurface processes. To better predict and pre-empt potential streamflow droughts, the relationship between recharge deficitand streamflow response needs to be studied more systematically. We present amodel experiment with a combined SVAT (soil-vegetation-atmosphere transfer)and conceptual groundwater model that applies a novel set of recharge scenariosto simulate preceding recharge drought events in humid-temperate catchmentswith different dominant aquifer types. The scenarios are based on (i) the permu-tation of historical time series and (ii) a modified time series to match an extreme(50-year) drought event. Time scales of aquifer-specific sensitivity to droughtvaried among aquifer types with catchments with karstic and fractured aquifersshowing more short-term sensitivity and porous and complex aquifers showingmore long-term sensitivity. This sensitivity of drought deficit and recovery timecorrelates with recharge age in baseflow, which is traceable with the groundwa-ter model and may provide a suitable measure of potential drought severity andan additional value for drought early warning systems. Different functional re-lationships between preceding recharge deficit and baseflow response in specificdrought years provided evidence that the aquifer-specific sensitivity to stream-flow drought can be classified as more climate- or more catchment-controlled.

33

Integrated drought monitoring systems – a tool fordrought research and practical applications

Miroslav Trnka1,2,mirek_trnka@yahoo.com

1 Institute of Agriculture Systems and Bioclimatology, Mendel University in Brno, CzechRepublic

2 Global Change Research Centre, Academy of Sciences of the Czech Republic

34

Characterisation of groundwater drought

John P. Bloomfield1, Ben P. Marchant1

jpb@bgs.ac.uk

1 British Geological Survey

Groundwater drought, typically expressed through anomalously low groundwa-ter levels or reduced groundwater discharge, is commonly conceptualised as atype of hydrological drought. Groundwater droughts can impact adversely onwater resources such as public water supply or water for industry and agri-cultural irrigation, as well as effecting groundwater discharge to groundwater-dependent surface waters and ecosystems. During the early stages of a ground-water drought, as deficits are developing in surface water and unsaturated zonestores, groundwater sources can provide relatively resilient water supplies andwill sustain surface flows through groundwater baseflow. Conversely, groundwa-ter may be highly susceptible to relatively persistent or prolonged meteorologicaldroughts, because, compared with surface water resources, groundwater storagemay take significantly longer to be replenished and recover as a drought beginsto break.

Like other forms of hydrological drought, the spatio-temporal response ofgroundwater droughts to precipitation deficits can be a function of catchmentcharacteristics and in particular it may be a function of aquifer distributionand spatial variations in aquifer characteristics. Aquifers typically act as lowpass filters, damping out high frequency components of driving meteorologysignals giving rise to longer term ‘memory’ or autocorrelation in groundwaterlevel hydrographs. Here we explore some of the implications of ‘memory’ orautocorrelation in groundwater levels for the characterisation of groundwaterdroughts.

The use of a standardisation method, the normal scores transform, a non-parametric normalization of data that assigns a value to observations based ontheir rank within a dataset, to index groundwater levels will be discussed. Specif-ically, we will show how the Standardised Groundwater level Index, SGI (Bloom-field & Marchant, 2013), scales with the Standardised Precipitation Index, SPI,

35

dependent on the autocorrelation structure of the normalised groundwater hy-drographs and the accumulation period of the SPI. Evidence for catchment andaquifer properties controls on the autocorrelation structure of SGI time serieswill be presented and discussed critically.

Using SGI hydrographs for 74 sites from three Principal Aquifers across Lin-colnshire, UK, we will describe how a combination of Cluster Analysis (CA)and Principal Component Analysis (PCA) have been used to regionalise thegroundwater drought response to meteorological droughts across the region forthe period 1983 to 2012. Contrasts in drought duration and magnitude betweenobservation sites in each of the six clusters identified will be analysed in termsof aquifer type and hydrogeological characteristics. In addition, the contrastingresponse of groundwater levels in observations boreholes in each of the six clus-ters to three major, multi-annual droughts will be explored in the context of theautocorrelation structure of individual SGI and mean cluster SGI hydrographs.Finally, some of the implications of the groundwater drought regionalisationresults for groundwater drought monitoring and warning will be discussed.

Using SGI hydrographs for 74 sites from three Principal Aquifers across Lin-colnshire, UK, we will describe how a combination of Cluster Analysis (CA)and Principal Component Analysis (PCA) have been used to regionalise thegroundwater drought response to meteorological droughts across the region forthe period 1983 to 2012. Contrasts in drought duration and magnitude betweenobservation sites in each of the six clusters identified will be analysed in termsof aquifer type and hydrogeological characteristics. In addition, the contrastingresponse of groundwater levels in observations boreholes in each of the six clus-ters to three major, multi-annual droughts will be explored in the context of theautocorrelation structure of individual SGI and mean cluster SGI hydrographs.Finally, some of the implications of the groundwater drought regionalisationresults for groundwater drought monitoring and warning will be discussed.

36

An alternative definition and characterisation of droughttermination

Simon Parry1, Christel Prudhomme2, Robert Wilby3, Paul Wood3

spar@ceh.ac.uk

1 Centre for Ecology & Hydrology and Loughborough University, UK2 Centre for Ecology & Hydrology, UK

3 Loughborough University, UK

The tendency for droughts around the world to terminate abruptly has beenfrequently observed in the historical record. The 2010-12 drought in the UKwas one of the most severe in the last 100 years, but the wettest nine-monthperiod for England & Wales in almost 250 years terminated the drought spec-tacularly. There has been a much greater focus in the literature on the onsetand characterisation of drought, whilst termination has been comparatively ne-glected. However, the end of a drought is the phase in which water resourcemanagers are most interested. This is the time that water is most scarce anddifficult decisions must be made about augmenting supply or limiting demand.More information on the likelihood of an upcoming terminal phase would aid thedecision-making of water managers, and an improved understanding of droughttermination could enhance drought monitoring and forecasting capabilities.

The existing body of research that explicitly considers drought terminationhas focussed on the amount of rainfall required to terminate drought, and thelikelihood of its occurrence. These are important questions, but few have at-tempted to characterise the terminal phase of a drought itself before consideringthe amount of rainfall required for this to occur. This may be partly explainedby the existing definition of drought termination, which has tended to focuson a specific point in time at the end of a drought when conditions have re-turned to ’normal’. This point is often calculated as the termination criteria of adrought index, but this theoretical instantaneous switch from drought to normalconditions understates the magnitude of change that can occur at the end of adrought. This existing approach would classify the often incredibly wet condi-tions of termination within the period of drought itself, since the terminationcriteria are not satisfied until wetness persists for a given duration. Hydrolog-

37

ical considerations of drought termination could learn much from the focus onrecovery that is prevalent in ecological science.

An alternative definition of drought termination is proposed which definesthe ’termination’, a period of recovery separating drought from normal condi-tions. Once this transition phase has been delineated, it can be described by anumber of indices including duration, abruptness and seasonality. These met-rics have been applied to the different elements of the hydrological cycle: rain-fall data as an indicator of meteorological drought termination; river flow datafor hydrological drought termination; and groundwater data for water resourcesdrought termination. This allows the investigation of the propagation of droughttermination through the hydrological cycle, a topic that has been neglected inthe existing literature. The terminal phases of a number of important histor-ical droughts in the UK have been characterised, allowing the spatio-temporalvariability in drought termination to be assessed. Indices for the 2012 droughttermination indicate that catchments in central and southern England recoveredabruptly, a finding that is consistent with previous analyses. Groundwater levelsalso generally returned to normal conditions remarkably rapidly, although hy-drological and groundwater drought termination varied according to catchmentand aquifer characteristics.

38

Linking drought indicators with impacts: Insights from acase study in Germany and the UK

Sophie Bachmair1, Kerstin Stahl2, Jamie Hannaford2, Lucy J. Barker2, IreneKohn1, Cecilia Svensson2, Maliko Tanguy2

sophie.bachmair@hydrology.uni-freiburg.de1 Chair of Hydrology, University of Freiburg

2 Centre for Ecology and Hydrology

Drought is a complex natural hazard with severe environmental and socio-economic impacts. Although little can be done in the short term to preventa drought, actions can be taken to reduce the vulnerability to drought, in-cluding the development of drought monitoring and early warning (M&EW)systems. To improve drought M&EW systems we need to better understandthe link between physical drought indicators and impacts on the environment,the society, and the economy. There is a vast range of published drought in-dicators available, but very little consensus on the most appropriate variablesand indicators for M&EW; in part this is because indicators have rarely beenlinked to observed impacts, so the meaningfulness of hydro-climatic indicatorsis typically far from clear. What does a certain SPI or streamflow indicatorvalue actually mean regarding the occurrence of drought impacts? Answeringsuch questions is a challenging task due to the complexity of how a prolongedprecipitation deficit propagates though the hydrological cycle and interacts withenvironmental and socio-economic factors. There have been very few attemptsto systematically characterize the relationship between drought indicators andimpacts owing to the sparse and patchy information on drought impacts. TheBelmont Forum funded project ’DrIVER’ (Drought Impacts and Vulnerabilitythresholds in monitoring and Early warning Research) is aiming to fill this gapby conducting analysis of the indicator-to-impact link in Europe, North Americaand Australia. This study is a first step on the road to exploring indicator-impactlinks in Europe for the DrIVER project, using the newly established EuropeanDrought Impact Report Inventory (EDII). The aim is to explore the link be-tween drought impacts and different commonly used drought indicators for twocountries well covered in the EDII: Germany and the UK. Specifically, we in-vestigate which drought indicators best explain impact occurrence, and whether

39

this depends on impact type and/or drought event. The focus of the study is ondata visualization and correlation analysis based on coarse-scale, major socio-economic regions (NUTS1 level), highlighting differences and commonalities inindicator-impact patterns between Germany and the UK. First results revealinteresting insights into the non-trivial relationship between drought indicatorsand impacts. The most suitable indicators for explaining drought impact occur-rence are SPI/SPEI for intermediate accumulation periods but there is variationamong events and geographies. This calls for more research on indicator-impactassessment as well as for further developing impact inventories. Next steps com-prise augmenting EDII data for drought events currently not well covered andapplying a similar methodology to further study areas in Europe and, eventually,in North America and Australia.

40

Influence of seasonal snowpack on summer low flows

Michal Jenicek1, Jan Seibert2, Massimiliano Zappa3, Maria Staudinger2,Tobias Jonas4

michal.jenicek@natur.cuni.cz

1 Charles University in Prague, Faculty of Science, Department of Physical Geography andGeoecology

2 University of Zurich, Department of Geography3 Swiss Federal Institute for Forest, Snow and Landscape Research WSL

4 WSL Institute for Snow and Avalanche Research SLF

The expected increase of air temperature could increase the ratio of liquid pre-cipitation during winter and, thus, decrease the snow accumulations especiallyin mid-elevation mountain ranges across Europe. The decrease of snow accu-mulation will affect groundwater recharge during spring and might cause lowstreamflow values in the subsequent summer period.

We assessed existing relations and trends in time series (1971-2012) of pre-cipitation, snow water equivalent and runoff in 25 alpine and pre-alpine basinsin Switzerland (basin area from 30 to 350 km2). The aims were 1) to quantifythe influence of winter and spring snow conditions on summer low flows and 2)to determine how long is the “memory effect” of basins in terms of the influenceof snowpack on stream runoff.

We tested different indices in order to find the best predictor of future summerlow flows (expressed as 7-day runoff minimum). We tested the maximum of SWE(snow water equivalent) in spring period, the sum of winter precipitation, thesum of precipitation during n days prior to the occurrence of minimum runoff,and groundwater storage estimated using a conceptual catchment model (HBV).The dependence of maximum SWE on the number of days with runoff below acertain threshold and the dependence of maximum SWE on Julian date of runoffminimum were also tested. All relations were assessed both for the whole summerperiod and for individual months (May – October).

The results indicated that the relations between seasonal snowpack and re-spective summer runoff were more complex than expected. There was a signifi-cant dependence of SWE maximum on runoff minimum in high-elevation basinswith high amount of precipitation stored in the seasonal snowpack. However,

41

only a minor portion of low flows variability was explained by the SWE (up to50%). Moreover, the influence of snow on summer low flows was significant onlyfrom May to August. There were only weak correlations in case of low flowsoccurring in September and October. In mid-elevation basins the correlationsare rather poor. On the other hand, low flows are occurring later in the year incase of above average snow accumulations and the number of days below certainrunoff threshold is lower as well.

The results showed a strong dependence of runoff minimum and the sum ofprecipitation n days prior the day with runoff minimum. The strongest correla-tions were achieved in case of accumulated precipitation 60 days prior the runoffminimum in specific month (explaining up to 80% of the variability). Despitesome influence of the seasonal snowpack on summer low flows we can concludethat rainfall conditions during spring and summer period are more importantfor low flows occurrence than snow accumulation during the preceding winter.

42

Relationship between groundwater level and NDVI:summer drought development analysed by in-situ and

MODIS-derived data

Urszula Somorowska1

usomorow@uw.edu.pl

1 University of Warsaw, Faculty of Geography and Regional Studies

In recent decades severity of droughts is reported to be an emerging issue acrossthe world, including Europe. Prolonged dry periods in summer lead to vege-tation water stress, sometimes of a flash nature. Besides climate, the physicalcatchment structure is considered an important control of hydrological droughtdevelopment although it is not always sufficiently represented in models (VanLanen et al., 2013). A range of different single or combined indicators is alreadyin use to detect and monitor droughts. However, further research is still neededto account for spatial heterogeneity and variability in the controls on water fluxesacross different scales (Asbjornsen et al., 2011). In this study, the response ofvegetation is evaluated by correlating a drought index characterizing ground-water system with the normalized difference vegetation index (NDVI), generallyconsidered as an efficient indicator of terrestrial vegetation productivity and sen-sitive to soil moisture fluctuations (Peled et al., 2010). The vegetation-subsurfacewater storage dependencies are identified at sites representing different ecosys-tems (forests and grassland) with limited human impact. They are located ina lowland watershed in central Poland (approximately 460 sq. km) in the UN-ESCO Biosphere Reserve (N52◦15’– N52◦24’ and E20◦15’–E20◦57’). Questionsinclude: (1) how shallow subsurface water storage is depleting in summer inrespond to meteorological drought at sites representing different ecosystems,(2) is there a signature of the vegetation conditions in soil moisture/shallowgroundwater indices; are they significantly correlated, (3) how sensitive are dif-ferent ecosystems to drought propagating through the subsurface. Time seriesof decreasing groundwater levels during growing season (April-September) wereused to detect classical rainfall deficit droughts. Regression model was fittedto quantify relation between NDVI and groundwater. It was evaluated if timelag is present between the starting of a water shortage and the signal of its

43

consequences, in this case by a decrease of the NDVI. Data comprise: (1) shal-low groundwater level of an unconfined aquifer monitored by 56 observationboreholes, supported by the temporal soil moisture measurements in 14 soil pro-files, (2) weather radar images at 1km resolution available for years 2003-2008,including years with extreme droughts, (3) normalized difference vegetation in-dex (NDVI), acquired from the MODIS Global Project which were temporallysmoothed by a non-linear function to reduce noise and possible residual cloudeffects. Results show that vegetation depends highly on a shallow subsurfacestorage.

References:Asbjornsen H., Goldsmith G.R., Alvarado-Barrientos M.S., Rebel K., Van Osch

F.P., Rietkerk M., Chen J., Gotsch S., Tobón C., Geissert D.R., Gómez-Tagle A.,Vache K., Dawson T.E.. Ecohydrological advances and applications in plant-waterrelations research: a review. Journal of Plant Ecology, 4, 1-2, 3-22, 2011.

Peled, E., Dutra, E., Viterbo, P., and Angert, A.: Technical Note: Comparingand ranking soil drought indices performance over Europe, through remote-sensing ofvegetation, Hydrol. Earth Syst. Sci., 14, 271-277, 2010.

Van Lanen, H. A. J., Wanders, N., Tallaksen, L. M., and Van Loon, A. F.: Hydro-logical drought across the world: impact of climate and physical catchment structure,Hydrol. Earth Syst. Sci., 17, 1715-1732, 2013.

44

Differentiating between influencing factors land use andclimate to assess drought effects on groundwater recharge

in a temperate context

Boud Verbeiren1, Marijke Huysmans1, Sven Vanderhaegen2, Frank Canters2, KlaartjeVerbeeck3, Guy Engelen3, Ingrid Jacquemin4, Bernard Tychon4, George Tsakiris5,

Harris Vangelis5, Okke Batelaan1

bverbeir@vub.ac.be

1 Vrije Universiteit Brussel (VUB), Dept. of Hydrology & Hydraulic Engineering, Brussels,Belgium

2 Vrije Universiteit Brussel (VUB), Dept. of Geography, Brussels, Belgium3 Unit Ruimtelijke Milieuaspecten, Vlaamse Instelling voor Technologisch Onderzoek

(VITO), Belgium4 Groupe Agrométéorologie, Département des sciences et gestion de l’environnement,

Université de Liège (Ulg), Belgium5 Centre for the Assessment of Natural Hazards and Proactive Planning, National Technical

University of Athens, Greece

Reliable estimation of groundwater recharge in space and time is of utmost im-portance for sustainable management of groundwater resources. Groundwaterrecharge forms the main source for replenishing aquifers. The main factors influ-encing groundwater recharge are the soil and topographic characteristics, landuse and climate. While the first two influencing factors are relatively static,the latter two are (highly) dynamic. Differentiating between the contributionsof each of these influencing factors to groundwater recharge is a challengingbut important task. Especially in case the main goal is to assess the effects ofdrought on groundwater recharge. On the one hand, the occurrence of meteoro-logical drought events is likely to cause direct, potentially deteriorating, effectson groundwater recharge. On the other hand, this is also the case for on-goingland-use dynamics such as extensive urbanisation. The presented methodologyaims at distinguishing in space and time between climate (drought-related) andland-use (human) effects, enabling to assess the effects of drought on groundwa-ter recharge in a temperate context. The physically-based water balance modelWetSpass is used to calculate groundwater recharge in a distributed way (spaceand time). Land-use dynamics are accounted for by using a consistent timeseriesof land-use (1990, 2000, 2006, and 2013) and fraction (vegetated, impervious,

45

bare & open water) maps. To account for climate variability a distributed me-teorological monthly timeseries for around 30 years is considered. A combineddrought index approach (RDI, SPI, sc PDSI) is used to identify meteorologicaldrought events during this period. WetSpass simulations are used to assess theweight of the influencing factors ‘land use’ and ‘climate’ with respect to droughteffects on the recharge timeseries. Hereto WetSpass is run several times with dif-ferent land-use input, while the dynamic meteorological timeseries (1980-2013)is considered for every scenario. Three simulation runs are used: (1) static land-use 1990, representing the historical maximum recharge conditions situation, (2)static land-use 2013, representing the current minimum recharge conditions and(3) dynamic land use 1980-2013, representing the combined land-use/climaterecharge conditions. The results of these three WetSpass simulations enable toassess the ‘land-use effect’ on groundwater recharge for each monthly timestep.As a consequence, the complementary ‘climate effect’ can be derived as well.The latter is important to identify potential effect of drought on groundwa-ter recharge, which is the main aim in this study. Important to note is that theWetSpass simulations also enable to assess spatial effects. This research is fundedwithin the SSD Programme of the Belgian Science Policy Office (BELSPO).

46

Reconstruction of meteorological droughts in France since1871 through the probabilistic downscaling of a global

atmospheric reanalysis

Laurie Caillouet1, Jean-Philippe Vidal1, Eric Sauquet1, Muriel Haond2

laurie.caillouet@irstea.fr

1 Irstea, UR HHLY Hydrology and Hydraulics Research Unit2 Compagnie Nationale du Rhône (CNR)

Hydroclimate projections for the 21st century agree on an increase in low flowseverity that could undermine current water management practice and requiredrastic measures for adapting water uses and for sharing resources among differ-ent economic sectors (irrigation, hydropower production, etc.). Understandingdroughts and low flows is all the more crucial in this current context of adap-tation to climate change. A deep knowledge of the major historical droughtsin France constitutes a perfect framework to assess the recent observed eventsbut also future events projected by climate change impact studies. Indeed, afundamental way of adapting to climate change tomorrow is to adapt to climatevariability today.

This work proposes a reconstruction of precipitation fields in France overthe last century built on the NOAA 20th century global atmospheric reanalysis(20CR). This reanalysis provides for the first time an evolution of atmosphericcirculation at high temporal resolution since 1871, assimilating surface pressureobservations only. The coarse spatial resolution of 20CR however prevents usingdirectly surface variables like precipitation at the scale of a French catchment.

The Sandhy (Stepwise ANalogue Downscaling method for Hydrology) statis-tical downscaling method is used here to bridge the scale gap between 20CR andprecipitation relevant for catchment-scale hydrology. This method is based onthe idea that similar large scale circulation patterns (predictors) lead to similarlocal precipitation (predictand). For a given target large scale situation, thedates with the most similar large scale situations in an archive are identified,and the corresponding local scale observations are selected as the targeted localscale predictions. For each target date, Sandhy gives an ensemble of 125 analoguedates and corresponding precipitation values, taking into account equifinality in

47

the optimization of predictor geographical domains. Such an optimization ismoreover done independently for 608 climatically homogeneous zones coveringFrance. Sandhy draws here analogue dates from an archive covering the 1958-2008 period for which local precipitation is also available through the Safrannear-surface reanalysis. It is then applied to target situations given by 20CRover the whole 1871-2012 period to derive an ensemble of 125 equally plausiblegridded precipitation time series.

Precipitation data are then monthly aggregated for each time series to com-pute the Standardized Precipitation Index (SPI) at different time scales andtherefore characterize meteorological drought events in a probabilistic way. TheSPI critically allows for comparing historical events and their spatio-temporaldevelopment. Results allow us to identify severe events for periods with fewavailable precipitation observations, such as the record-breaking 1921 drought.Such an event – much more severe than the recently experienced ones like 1976 or2003 – may thus serve as a benchmark reference event for adaptation purposes.

48

Tree rings as a tool for reconstructing historic droughts innortheastern Austria

Sandra Karanitsch-Ackerl1, Michael Grabner1, Franz Holawe2, Gregor Laaha1

sandra.karanitsch@boku.ac.at1 University of Natural Resources and Life Sciences Vienna (BOKU)

2 University of Vienna

The northeastern part of Austria, the Weinviertel region, belongs to the warmestand driest of the country’s areas with a mean annual air temperature of 9.3 ◦Cand a mean precipitation of 485 mm per year (Retz 1971-2010).

The region is predominantly used for agriculture. Forests cover only ap-proximately 15% of the area. Oak (mainly Quercus robur and Q. petraea) isthe natural dominant species, but also planted pine species (Pinus nigra and P.sylvestris) grow here.

Water availability is a crucial factor for tree growth in this region. Annualincrement is highly dependent on the amount of precipitation – especially duringspring and summer. A pronouncedly dry vegetation period (as for example in2003) normally results in a very narrow annual ring. This circumstance makestree rings a valuable proxy for drought reconstruction.

During the Sparkling Science project WALD-WASSER-WEINviertel (SPA03-015) funded by the Austrian Federal Ministry of Science, Research and Economy,multi-century chronologies for oak (almost 800 years – 1216 to 2011) and pine(more than 500 years – 1472 to 2011) each consisting of more than 400 sam-ples from living trees as well as from historic buildings were set up. For bothtree species, three separate chronologies – for ring width, earlywood width andlatewood width – were constructed.

They can now serve as a tool for reconstructing regional droughts of the pastcenturies.

For this purpose, first of all, Spearman correlations of the chronologies with(1) monthly precipitation, (2) monthly mean temperature and (3) a self-madedrought index based on the Walter-Lieth climate classification as well as with(4) the climatic water balance calculated with potential evapotranspiration ac-cording to Thornthwaite for the stations Vienna, Retz and Brno (1897-2010)were investigated.

49

First results for ring width are promising. The pine chronology correlatessignificantly with spring and summer months. The highest correlation coef-ficients (approximately 0.7) are achieved with precipitation in Vienna for thesix-months-period March to August. The values for oak are generally lower andsomewhat different in seasonal timing.

Further investigations of the climate-growth-relationships are necessary andcurrently in progress. The goal of the study is to establish a model for recon-structing dry periods of the past using the differing drought reactions of oak andpine as well as earlywood width and particularly latewood width, which alsocorrelates with summer river runoff.

50

Developing a multi-disciplinary inventory of UK droughtsfrom the late 19th century

Jamie Hannaford1, John Bloomfield2, Ian Holman3, Bettina Lange4, RalphLedbetter5, Mark McCarthy6, Tony McEnery7, Rebecca Pearce8, Steven

Wade6

jaha@ceh.ac.uk

1 Centre for Ecology and Hydrology, United Kingdom2 British Geological Survey, United Kingdom

3 Cranfield University, United Kingdom4 University of Oxford, United Kingdom5 HR Wallingford Ltd, United Kingdom

6 Met Office, United Kingdom7 University of Lancaster, United Kingdom

8 University of Exeter, United Kingdom

An understanding of past droughts is essential for contemporary and future watermanagement. Research on the meteorological drivers of past droughts, and howthese propagated through hydrological and water resources systems, is vital fordeveloping improved drought management practices. However, typically manyhydrological datasets are relatively short, and only shed light on a handful ofmajor drought episodes in the latter decades of the 20th Century. Considerableeffort has been invested in analysing historic droughts in the UK using recon-struction techniques, with a number of major drought episodes documented sincethe late 19th Century. However, a limitation of previous research on UK historicdroughts has been their focus on hydro-meteorological characteristics, and thereis currently no comprehensive understanding, inventory or typology of droughtsfor the UK that integrates their socio-economic drivers and impacts with naturalscience based characterisations of drought. This is a major gap in research, asdroughts are not simply natural hazards but can also be the consequence of spe-cific economic, social and legal practices, such as water consumption practicesand abstraction licensing. Moreover, episodes of drought are rarely experiencedin the same way by all communities or sectors. Improved drought managementrequires a more integrated perspective informed by an understanding of the linksbetween environmental and social systems during past droughts.

51

The ‘Historic Droughts’ project is a joint UK Research Council funded projectthat will develop a comprehensive inventory of past droughts in the UK. Atruly multi-disciplinary perspective is taken in order to develop an inventorythat assembles accounts of drought events from the perspectives of meteorology,hydrology, economics, law, sociology and history. The inventory will includetimelines of past droughts as they have manifested in different sectors, such ashydrometeorology and the environment, water resources management, agricul-ture, regulation and policy and citizens’ perceptions.

The project embraces diverse data sources and novel methodologies to recon-struct past drought characteristics in each of these sectors, including, for exam-ple: hydrological modelling reconstructions to extend river flow and groundwaterrecords back to the late 19th Century; assessment of water resources manage-ment interventions in past droughts using water company records; reconstructionof agricultural impacts through proxies, e.g. crop prices; development of time-lines of regulatory reform milestones through records of political debate in theHouse of Commons; collation of oral histories based on interviews in communi-ties affected by major past droughts, and with water professionals involved intheir management; computer-aided discourse analysis of newspapers from the19th and 20th centuries; and analysis of social media data from 2010 – 2012, thefirst drought in the age of twitter.

The drought inventory is designed to provide a knowledge-base of past droughtoccurrence that will then be used within the ‘Historic Droughts’ project to de-velop a systems-based understanding of drought, reflecting the fact that droughtsare complex hazards, shaped by manifold human and natural drivers, impactsand their interactions. By learning from the past, a better understanding of co-evolving social and natural systems can inform improvements in the managementand communicating of droughts in future.

52

The analysis of storage discharge relationship for theidentification of low flows on selected watersheds in the

Czech republic

Petr Maca1, Adam Vizina2, Stanislav Horacek2, Martin Hanel2, LadislavKasparek2, Pavel Pech1

maca@fzp.czu.cz

1 Dept. of Water Resources and Environmental modeling, FES, CULS, Prague, CzechRepublic

2 T. G. Masaryk Water Research Institute, p.r.i., Prague, Czech Republic

We analyzed the daily hydrological rainfall runoff data of 13 watersheds in theCzech republic. The main assumption of this contribution is that the low flowscan be described by a single reservoir controlled by the constitutional storagedischarge relationship. The proposed analysis combines the identification ofparameters of selected storage discharge relationship using the recession limbsanalysis, the global evolutionary optimization algorithm and lumped concep-tual model Bilan. Three different smoothing methods were applied on runoffdata for the selection of recessions segments of hydrographs. Their influenceon the estimation of baseflow parameters is further discussed. Two differentconstitutional storage discharge relationships were identified for those datasetsand compared. The identifications of studied baseflow parameters were madeusing two methods. The first was based on the original drought flow analysis ofBrutsaert et Nieber [1], the second method applied the global single objectiveevolutionary optimization algorithm SCDE for the estimation of parameters oftested constitutional relationships controlling the low flows. The SCDE opti-mization algorithm combines the original work of Storn and Price [2] with theshuffled complex evolution [3]. Based on the results of recession curve analysisthe ASVH tool for the identification of parameters of tested storage discharge re-lationship was compiled. The ASVH enables the ensemble optimization run andprovides the information about uncertainty in parameter estimation. Finallythe parameters of lumped model Bilan were identified using the newly acquiredinformation about parameters, which controls the low flows simulation and isobtained from ASVH analysis. The results show that the proposed analysis im-

53

proves the identification of parameters of lumped hydrological model Bilan. Thesuggested analysis can be easily adopted for any hydrological model.

References:[1] Brutsaert, W; Nieber, J. L. (1977). Regionalized drought flow hydrographs from

a mature glaciated plateau. Water Resources Research, 13: 637–643.[2] Storn, R.; Price, K. (1997). Differential evolution - a simple and efficient heuris-

tic for global optimization over continuous spaces. Journal of Global Optimization 11:341–359.

[3] Duan, Q; Sorooshian, S; Gupta, V. K. (1994). Optimal use of the SCE-UAglobal optimization method for calibrating watershed models, Journal of Hydrology,158, 3–4:265-284,

54

Current and future drought vulnerability assessment inthe Jucar River basin (Spain)

David Haro1, Abel Solera1, Javier Paredes-Arquiola1, Joaquín Andreu1

David.Haro.Monteagudo@gmail.com

1 Institute of Water and Environmental Engineering, Universitat Politècnica de València

Drought occurrence and its related impacts are a major concern in many riverbasins throughout the world. In the coming times, water managers will confrontsevere uncertainties within the decision making process, both in the short term(management and operation) and long term (planning), in any water resourcessystem where droughts are very frequent and where water resources are under amassive use. In general, vulnerability is used to characterize the performance ofthe system and it is a good indicator of the most likely failures. This may helpmanagers especially in the planning phase, in which new measures both technicaland operational may be devised to reduce the risk of suffering the effects of adry spell either in the short or long term.

Decision support tools are a great ally of water managers, especially in situ-ations of high water stress and hydrologic uncertainty, allowing them to in-deepanalyze the system and finding the best measures to minimize the risk of asystem failure. The use of such systems allows developing and using real timemanagement models able to assess the vulnerability and risk of drought, and theeffectiveness of proactive and reactive measures applied on regular basis for themanagement of river basins. They also represent a powerful tool for participa-tory processes since the different stakeholders involved in drought planning havea chance to develop and understanding of one another’s various points of view,and to generate collaborative solutions.

The Jucar River basin is one of the most vulnerable areas to drought of thewestern Mediterranean region due to climate characteristics (mostly semiarid),high water exploitation indexes (water scarcity), and very high space and timevariability of precipitation (leading to a highly seasonal, and inter-annual vari-ability in the river flows). Hydrological droughts in this region are frequent,and they can be very intense and with duration of years (frequently more than3 years). For these reasons, careful risk and vulnerability assessments must be

55

carried out to find and apply the appropriate measures to reduce them. Climatein the Jucar River basin is characterized by great variability, and climate projec-tions only reveal that future resources generation will be characterized by greatuncertainty.

In this article, we use a Monte Carlo approach to evaluate current and fu-ture vulnerability to drought in the Jucar River basin. First, we will assess theexisting drought vulnerability under the present climatic conditions with a sim-ulation model feed with multiple equiprobable series. Since it is not possible toknow what will be the future operation of the system, its vulnerability needs tobe approached from a different perspective. We will use an optimization modelinstead to obtain the minimum expected vulnerability of the system. This canhelp managers to develop future operation rules and enhance participatory pro-cesses to show stakeholders what are the least problems they may confront inthe future.

56

Droughts in the 21st century: a global perspective

Christel Prudhomme1,chrp@ceh.ac.uk

1 Centre for Ecology & Hydrology, United Kingdom

57

Future hydrological drought on a global and continentalscale: overview and outlook

Henny A.J. van Lanen1, Niko Wanders2, Marjolein H.J. van Huijgevoort1,3,Yoshihide Wada2, Marcel A. A. Alderlieste4

henny.vanlanen@wur.nl

1 Hydrology and Quantitative Water Management Group, Wageningen University2 Department of Physical Geography, Faculty of Geosciences, Utrecht University

3 School of Geosciences, University of Aberdeen, 4 Hydrologic, Utrecht

Drought likely will become more severe in multiple regions across the world,although confidence in projections is still low. Most studies on future droughtusually address meteorological drought (Standardised Precipitation Index) orsoil moisture drought (Palmer Drought SI, Soil Moisture Anomaly). So far, onlya few studies dealt with future drought in groundwater or river flow (hydrologicaldrought) on a continental scale or on a global scale. This paper aims to presentan overview on hydrological drought on large scales in the 21st century undernatural conditions and then to provide an outlook on how to proceed.

The first studies on large-scale hydrological future hydrological drought usedprojected annual runoff from Global Circulation Models (GCMs). The outcomewas very much connected to change in the driving force, in particular the an-nual precipitation. Trend studies on historic drought showed that projections ofannual total flow only is too broad, because relevant intra-year changes occur.Moreover, the spatial scale of the GCMs was considered to be too coarse.

In the last 5-10 years, outcome from GCMs was downscaled and bias cor-rected and then used as driving force for off-line land surface models (LSMs) andglobal hydrological models (GHMs/GIMs). The spatial scale of the LSMs andGHMs/GIMs (50 km2) is more appropriate to contemplate differences in landsurface properties and hydrolog. processes can be implemented in more detail.

Studies in transects across Europe using three CMIP3 GCMs show that themulti-model mean annual river flow under natural conditions likely will decrease(max. 40-70%) with a higher probability by the end of the century. The noise(lack of skills of GHMs fed with GCM to reconstruct past simulated runoff)is rather large relative to climate change signal. The climate change - noise

58

ratio is much larger for drought characteristics. The average drought durationis projected to increase (B1 scenario: 100-200%, A2 scenario: 275-325%) by theend of the 21st century. A study with the same CMIP3 CGMs and a singleGHM for all climate regions across the globe confirm the increase in droughtduration. The increase in average duration is connected to a substantial lowerdrought frequency. The projected severity (aver. deficit volume) of the remainingdrought events will increase by 100-125% (A2 scenario, 2071-2100). In a finalstudy using CMIP3 GCMs, observed river flow from 41 river basins across theglobe was used to develop a methodology to select combinations of GCM-GHM(n=15) to reduce the noise in the assessment of future low flow and hydrologicaldrought. In about 65% of the river basins there were sufficient GCM-GHMcombinations (n>=5) to predict future hydrological drought, but in not morethat 25% of the river basins 10 or more combinations were available.

More recently, future hydrological was assessed by using five CMIP5 CGMs,seven GIMs and four emission scenarios (RCPs). Under the most extreme sce-nario (RCP8.5), the area in drought is projected to exceed 40% of the global areain almost 50% of the GCM-GIM combinations with a strong climate change –noise ratio. In a follow-up study with the same GCMs and RCPs, one singleGIM was used to explore hydrological drought in the 21st century that considersadaptation to future changes in hydrological regime (transient vs. non-transientvariable threshold method, VTMt vs. VTM). The global area in drought in2071-2100 is rather similar to 1971-2000 when drought was estimated using agradually changing hydrological regime (VTMt). In about 25% of the globe,both the average drought duration and the deficit volume is projected to increase(RCP8.5) by the end of the 21st century (VTMt), whereas both characteristicswill increase in over 60% of the globe (VTM).

Based upon the experiences with the large-scale studies using downscaledand bias-corrected CMIP3 and CMIP5 GCM outcome in LSMs, GHMs/GIMs,improvement of the assessment of future hydrol. will be discussed, which include:(i) role of catchment storage, (ii) hyper resolution modeling and parameterisa-tion, (iii) selection of models in a multi-model setting, (iv) definition of drought(transient def.), and (v) influence of human impacts (water use and reservoiroperation).

59

Hierarchy of climate and hydrological uncertainties in lowflow projections

Jean-Philippe Vidal1, Eric Sauquet1, Claire Magand2, Agnès Ducharne2,Benoît Hingray3

jean-philippe.vidal@irstea.fr

1 Irstea, UR HHLY, Hydrology-Hydraulics Research Unit2 METIS, UPMC/CNRS – UMR7619, Université Pierre et Marie Curie

3 CNRS/UJF-Grenoble 1/G-INP/IRD, LTHE UMR 5564

This works aims at providing a understanding of the different types of modellinguncertainties associated with future low-flow projections. Hydrological projec-tions come indeed at the bottom of the well-known cascade of uncertainty andtherefore potentially integrate all types of modeling uncertainties related to bothclimate and hydrological. The specificity of low flow characteristics is their tightdependency on catchment processes, suggesting a non negligible contribution ofthe hydrological modelling step to the overall uncertainty.

This study builds on an extensive top-down climate impact approach set upwithin the R2D2-2050 project. This integrated assessment project aims at in-forming water resource adaptation strategies for the Durance river basin locatedin the Southern French Alps, by confronting future hydroclimate projections toprospective scenarios of water demand and water use. Future water resourceestimates have been based on 11 runs from 4 GCMs from the ENSEMBLESproject under the A1B emissions scenario. These large-scale projections havebeen further statistically downscaled with 3 probabilistic methods based on thek-nearest neighbours resampling approach, leading to 330 spatially distributedclimate projections. Downscaled projections have then been used as forcings to6 diverse hydrological models, from global conceptual to physically-based fullydistributed, in order to derive transient hydrological projections from 1961 to2065 for up to 25 stations over the basin.

Most of the Durance subcatchments are under the influence of both theAlpine and Mediterranean climates, which leads to two distinct periods of lowflows, one in winter when precipitation is stored as snowpack and one in latesummer after the snowmelt. Analyses have therefore been applied separately to

60

the two periods to infer the future evolution of low flows characteristics and itsconsistency among the range of available projections. Several analysis of vari-ance (ANOVA) frameworks have further been applied to quantify the respectivecontributions of climate and hydrological modeling uncertainties on the resultsand their evolution over time.

Results first show contrasted evolutions according to the season and thecatchment, with little evolution in winter low flows for high-elevation catch-ments, and a dramatic decrease in summer low flows over the whole basin. Thecontribution of the hydrological modelling step to the overall uncertainty is more-over limited in winter, but high in summer with a pronounced increase over time.This last result critically suggests a divergence of climate change responses be-tween the hydrological models in terms of summer low flows, potentially due todifference in the implementation of snowmelt processes.

61

Impacts of Anthropogenic Emissions on Hydrologic Cycle

Venkatachalam Ramaswamy1

v.ramaswamy@noaa.gov

1 NOAA/ Geophysical Fluid Dynamics Laboratory, and Atmospheric and Oceanic SciencesProgram, Princeton University

We focus on the significance of anthropogenic emissions of long-lived and short-lived greenhouse gases and aerosols on the hydrologic cycle. Using the NOAA/GFDL global climate models developed for and applied in the Intergovernmen-tal Panel on Climate Change Fourth (2007) and Fifth (2013) Assessments, weinvestigate the perturbations induced by the human-induced emissions on theradiative and water balance of the planet, from the global to the regional spacescales. Of particular interest is the manner in which the external forcing agentscause changes in the surface energy balance, both in clear and cloudy skies,since the pre-industrial period and into the 21st Century. We differentiatebetween the simulated changes due to the anthropogenic greenhouse gas andaerosol emissions, and compare the climatic impacts with those obtained for thenatural drivers (e.g., unforced variability, solar and volcanic-induced changes).The resulting influences due to the radiative forcings on the atmosphere andprecipitation features are analyzed for the different continents, including thepropensity for the initiation of precipitation deficit situations. Where possible,comparisons are made with available observations. Besides the trends on themulti-decadal scales, higher resolution climate models are revealing characteris-tic regional features that are increasingly better simulated. The model-simulatedfeatures, trends and variations are being deployed in a number of impacts sectorsof interest to water resource managers and agriculture e.g., regional-scale precip-itation deficit investigations in southwest US and Asia, precipitation severity inEurope and North America in the 21st Century. Limitations in the applicabilityof the models, and how the current state of affair can be improved, will also bediscussed.

62

Are runoff changes in Europe attributable toanthropogenic climate change?

Lukas Gudmundsson1

lukas.gudmundsson@env.ethz.ch

1 ETH Zurich Institute for Atmospheric and Climate Science

Changes in terrestrial water availability can have pronounced consequences forecosystem processes, land-atmosphere interactions and water resources planning.Although previous studies have found changes in European streamflow with in-creasing drought in the south and wettening conditions in the north, there isto date no formal assessment of the significance of these changes. Similarly, theeffect of anthropogenic climate change on these patterns has not been assessed.This leaves many pending questions regarding the significance and causes ofchanges in terrestrial water availability in Europe. Here we investigate regionalpatterns in changes of water availability in Europe, using a newly assembledcollection of homogenized streamflow observations that combines data from theEuropean Water Archive (EWA) and the Global Runoff Data Centre (GRDC).Using detection and attribution techniques, emerging patterns of change arecompared to patterns derived from comprehensive Earth System Model simula-tions driven with natural forcing only, and to simulations forced with historicalchanges in anthropogenic greenhouse gas emissions. The results, consequently,allow clarifying whether regional water availability in Europe has changed sig-nificantly over the past decades and if detected changes are attributable to an-thropogenic greenhouse gas emissions.

63

Spatio-Temporal Drought Analysis Under VaryingClimatic Conditions

Marzena Osuch1, Joanna Doroszkiewicz1, Hadush Kidane Meresa1, RenataRomanowicz1

marz@igf.edu.pl

1 Institute of Geophysics Polish Academy of Sciences

The aim of the paper is an estimation of changes induced by climate variability onthe drought severity in Poland using the Standard Precipitation Index (SPI). Theclimatic variables are obtained from the ENSEMBLE project, in the form of timeseries of precipitation derived from six different RCM/GCMs: DMI HIRHAM5ARPEGE, SMHIRCA BCM, RM51 ARPEGE, MPI M REMO ECHAM5, KNMIRACMO2 ECHAM5 r3 and DMI HIRHAM5 BCM following A1B climate changescenario for the time periods: 1971-2000, 2021-2050 and 2071-2100. Monthlyprecipitation time-series were used for the estimation of Standard PrecipitationIndex (SPI) for multiple time scales (1-,3-, 6-, 9-, 12- and 24 months) and spatialdiscretisation of 25x25 km2 for the whole country. In the first stage the resultedSPI indices for the reference period (1971-2000) were compared with SPIs derivedon the basis of e-obs reanalysis for the same period. We also compare the resultsobtained with and without bias correction (quantile based mapping method) ofthe RCM/GCM outputs. Results indicate a large influence of RCM/GCMs andbias correction on the low flow indices obtained. Comparison of indices from theperiods 2021-2050 and 2071-2100 with the reference period 1971-2000 confirmsthe trends obtained by previous research, in the form of a projected increase offrequency and intensity of droughts. The projected changes in the near future arelarger than those in the far future. Apart from the choice the simulation modelalso the choice of the time-scale has a large impact on the results. The studyshows that SPI indices can be useful in spatio-temporal analysis of droughts inthe region.

64

Statistical Modeling of Low Flow Conditions based onClimatic Indicators

Anne Fangmann1, Uwe Haberlandt1

fangmann@iww.uni-hannover.de

1 Leibniz Universität Hannover Institut für Wasserwirtschaft, Hydrologie undlandwirtschaftlichen Wasserbau

Making use of regression-based approaches in climate change impact assessmentmay pose a practical alternative to the application of process-based hydrologicalmodels, especially with respect to low flow extremes. Extended durations andspatial dimensions allow for a quantitative assessment and exploitation of the in-terrelations between atmospheric driving forces and streamflow response duringdry periods, and eventually for the prognosis of future low flow conditions basedon climate model input. This study aims at using combinations of climatic indi-cators, quantifying a variety of meteorological drought characteristics, to modelspecific low flow indices, based solely on multiple linear regressions. The areaunder investigation is the federal state of Lower Saxony, Germany. Records ofdaily streamflow of variable length are available for 268 gauges, while climaticvariables, interpolated on a 1x1-km grid exist for the period 1951 to 2011. Thedaily time series pose the basis for calculation of a set of meteorological andhydrological indices, serving as regressors and regressands, respectively. Twoapproaches are being analyzed: a) a station-based approach, fitting a specificregression equation at each discharge gauge with sufficient record length, andb) a regional approach, enabling the estimation of low flow indices at ungaugedsites and stations with minor record length. The station-based procedure is usedfor estimation of annual low flow index values from annual meteorological condi-tions. Subsequent fitting of distribution functions to the estimated values allowsfor the assessment of return periods of the low flow indices. The regionalization,on the other hand, is designed to directly estimate the shapes of the distribu-tion functions by applying L-moment regressions, enabling a direct assessmentof specific index values for the return periods in demand.

65

Spatial and temporal modelling of UK groundwaterdroughts using empirical models

Ben Marchant1, John Bloomfield1, Jon MacKay1

benmarch@bgs.ac.uk

1 British Geological Survey

Groundwater droughts in the UK adversely affect water supplies to the public,industry and agriculture in addition to limiting groundwater discharge to surfacewaters and ecosystems. Groundwater drought events are spatially heterogeneousreflecting variations in rainfall and aquifer properties. Variations in factors suchas the volume of groundwater stores and the hydraulic conductivity of aquifersinfluence the local resilience of groundwater to meteorological drought and therate at which groundwater supplies recover once the meteorological drought haspassed.

A thorough understanding of historic drought episodes is valuable in fore-casting the evolution of future droughts, in predicting where the effects of thedrought will be most strongly felt and in targeting mitigation efforts. Thisunderstanding can be expressed via empirical models of the spatial and tempo-ral variation of groundwater levels. Given the spatial heterogeneity of droughtepisodes it is vital that these models reflect a wide range of locations.

Organisations such as the Environment Agency and the British GeologicalSurvey hold records of groundwater levels from thousands of UK boreholes.These data can potentially be used to calibrate empirical models. However thereare many challenges associated with such exercises. The groundwater recordsare of different durations. The groundwater levels are often observed at irregularintervals and incidents such as the 2001 foot and mouth outbreak prevent accessto many sites. This irregular sampling means that the records are not necessarilyfully representative of the groundwater level variation within the borehole and itimpedes the use of standard statistical time series models and the identificationof drought events.

We describe the development of a robust methodology to model the tempo-ral variation of groundwater levels in each borehole using linear mixed models(LMMs) which are estimated by residual maximum likelihood. These models are

66

more commonly associated with representations of spatial variation. The LMMsrepresent the temporal groundwater level variation in terms of fixed and ran-dom effects. The fixed effects account for the underlying variation – in our modelthis variation is driven by accumulated rainfall at the site. The accumulationperiod varies from site to site (reflecting variations in aquifer properties) and isautomatically selected via a series of likelihood ratio tests. The random effectsrepresent the temporal correlation amongst the data and allow us to accountfor the irregular sampling of the data. The approach is generally applicable andefficient and is hence suitable to be applied to the large number of availablegroundwater records.

We use the models to estimate measures of the duration and severity of UKdroughts and to determine the uncertainty of these estimates. We validate themodels against previous drought episodes (not used in the model calibration).Finally we consider how the model outputs from the different boreholes canbe interpolated to produce a continuous representation of both the spatial andtemporal variation of groundwater levels and groundwater drought status.

67

Spatial and temporal variability of droughts in Switzerland

Gaetano Di Franca1,2, Peter Molnar2, Paolo Burlando2, Antonio Cancelliere3

gaetano.difranca@gmail.com

1 University of Catania, Department of Civil Engineering and Architecture2 Dept. of Civil, Environmental and Geomatic Engineering ETH Zurich, Institute of

Environmental Engineering, Switzerland3 University of Catania, Department of Civil Engineering and Architecture, Italy

During the last decades, the Alpine region, and Switzerland in particular, hasexperienced several extreme meteorological events, including both droughts andfloods, which have led to severe damage and fatalities. Although Switzerlandis not an arid or semi-arid region, natural climatic variability makes it poten-tially prone to droughts which mostly affect agriculture and water supply forhydropower production. After the heat wave experienced in several Europeancountries in 2003, various areas of scientific research (meteorological, hydrologi-cal, agricultural, ecological, environmental, etc.) have focused on drought in theAlpine region, in order to assess whether it was a rare meteorological event orpossibly a signal of climate change.

In this work, spatial and temporal features of meteorological droughts inSwitzerland have been investigated through identification and quantification ofthe number of events, duration, severity and areal extent of droughts on griddedmonthly precipitation data for the entire country (2x2 km) during the period1961-2012. To this end, the Run method has been employed to probabilisticallycharacterize drought occurrences as well as to analyze their spatial variability.Spatial features have also been investigated by means of Principal ComponentsAnalysis (PCA) applied to Standardized Precipitation Index (SPI) series with12-month aggregated time scale, in order to detect areas with distinct droughtvariability. Furthermore, a probabilistic analysis of drought areal extent hasbeen carried out by applying an SPI-based procedure to derive Severity-Area-Frequency (SAF) curves.

Results reveal that Switzerland has experienced droughts in the last 50 yearswith prolonged dry spells and that there are regions in the south of Switzerlandwhere the occurrence of droughts seems to be more severe than in other regions.In particular, the application of the Run method shows that Ticino and Valais

68

are the most potentially prone drought regions, since accumulated deficit in pre-cipitation is significantly higher (up to two times) in comparison to other Swissregions. Anomalies in rainfall exhibit high spatial correlation, showing uniformsub-regions that have climatic similarities. These results are also confirmed byPCA. The SAF analysis indicates that precipitation anomalies have in fact af-fected large areas of Switzerland in the past, and not only in 2003 (e.g. January1962, August 1991).

This research provides the methodology for studying the drought phenomenonin the Alpine region. The results of the work show the importance and needof investigating droughts also in the Alpine region where the perception andawareness of this natural hazard is still lacking, while fast economic growth inagriculture, industry and energy production as well as changes in climate couldincrease drought risk in the future.

69

The PREMHYCE project: a comparative evaluation ofhydrological models for low-flow simulation and

forecasting on French catchments

Pierre Nicolle1, Raji Pushpalatha1, Charles Perrin1, Didier François2,Dominique Thiéry3, Thibault Mathevet4, Matthieu Le Lay5, François Besson5,

Jean-Michel Soubeyroux5, Christian Viel5, Fabienne Regimbeau5, VazkenAndréassian1, Pascal Maugis6, Bénédicte Augeard7, Emmanuel Morice8

pierre.nicolle@irstea.fr

1 Irstea, HBAN, Antony, France ; 2 Université de Lorraine, LOTERR ; 3 BRGM4 EDF-DTG ; 5 Météo-France, Direction de la Climatologie ; 6 IPSL, LSCE ; 7 ONEMA

8 Direction de l’eau et de la Biodiversité, Ministère de l’écologie, du développement durableet de l’énergie

Low-flow simulation and forecasting remains a difficult issue for hydrologicalmodellers, and intercomparison of approaches can be extremely instructive to as-sess existing low-flow prediction models and to develop more efficient operationaltools. This communication will present the main results of the PREMHYCEproject, a collaborative experiment conducted to compare low-flow simulationand forecasting models on catchments in France. Five hydrological models (fourlumped storage-type models and one distributed physically-oriented model) wereapplied within a common evaluation framework and assessed using a commonset of criteria. Two simple benchmarks describing the average streamflow vari-ability were used to set minimum levels of acceptability for model performancein simulation and forecasting modes.

Tests on 21 unregulated catchments were recently detailed by Nicolle et al.(2013). Results showed that, in simulation as well as in forecasting modes, all hy-drological models performed almost systematically better than the benchmarks.Although no single model outperformed all the others for all catchments andcriteria, a few models appeared more satisfactory than the others on average.In forecasting mode, we defined maximum useful forecasting lead times beyondwhich the model does not bring useful information compared to the benchmark.This maximum useful lead time logically varies among catchments, but also de-pends on the model used. Simple multi-model approaches that combine theoutputs of the five hydrological models were tested in an attempt to improve

70

simulation and forecasting efficiency. We found that on average the multi-modelapproach was more robust and can provide better performance than individualmodels.

The presentation will mainly focus on the complementary tests made on 11additional catchments regulated by dams or influenced by water withdrawals.The methodology applied to account for human influences will be explained.The main differences with the results obtained on unregulated catchments willbe presented, and we will discuss the operational perspectives of this project.

71

Trade-Offs between dynamical and statistical forecastingof the European 2003 Drought

Stephan Thober1, Luis Samaniego1, Rohini Kumar1

stephan.thober@ufz.de

1 Helmholtz Centre for Environmental Research - UFZ Permoserstr. 15, 4318

Agricultural droughts are considered among the most catastrophic natural dis-asters because they have the potential to diminish crop yields causing economicdamage or threatening the livelihood of societies. An agricultural drought oc-curs if a soil moisture index is falling below a threshold (typically, a quantilethat is equaled or not exceeded 20% of the time). State of the art drought mon-itoring and forecasting systems combine information from various observationaldata sources to provide the best possible estimate of current drought conditions.Additionally, some of these systems also incorporate seasonal meteorologicalforecasts to estimate future drought conditions. Meteorological forecasting skill,however, in particular that of precipitation, is limited to a few weeks because ofthe chaotic behavior of the atmosphere. One of the most important challengesin drought forecasting is to understand how the uncertainty in the atmosphericforcings (e.g., precipitation) is further propagated into hydrologic variables suchas soil moisture.

A large atmospheric ensemble has the potential to overcome the aforemen-tioned challenge. The North-American Multi-Model Ensemble (NMME) is thelatest collection of a multi-institutional seasonal forecasting ensemble and allowsto test the hypothesis that a large meteorological ensemble increases droughtforecasting skill at various lead times. In particular, it is investigated if the en-semble is already large enough for achieving a robust drought forecasting by sub-sequently selecting different sub-ensembles from it and evaluating their forecast-ing skill. Sub-ensemble selection methods are readily available and have alreadybeen successfully applied in the evaluation of ENSEMBLES regional climatemodels (RCMs) to reproduce observed extreme meteorological indices. Further-more, the forecast provided by these dynamical models is compared to that ofa simple statistical forecasting model. This statistical model uses geopotential

72

height and antecedent soil moisture conditions to provide future soil moistureestimates.

The aforementioned hypothesis is tested for the European 2003 droughtevent. The monthly NMME forecasts are bias corrected and downscaled todaily values to force the mesoscale hydrological model (mHM). The modelingperiod encompasses the period from January 1st to September 30th 2003. Thestudy area ranges roughly from 10 ◦W to 40 ◦E and 35 ◦N to 55 ◦N and coverslarge parts of the Pan-European domain.

Preliminary results indicate that the full NMME leads to more robust droughtstatistics than those obtained if only limited forecasts are considered. Based onthe fact that sub-ensemble selection has proven to be beneficial for increasingthe skill of RCMs to reproduce extreme indices, it is expected that a subset ofthe full NMME ensemble could be found that exhibits at least the same fidelityto forecast the European 2003 drought event as the full ensemble. Furthermore,it is expected that the statistical model is able to provide a more skillful soilmoisture forecast on short lead times (i.e. less than three months), but is out-performed by the forecast achieved with the dynamical models on longer leadtimes (i.e. six to nine months).

73

Investigating The Potential Contribution OfTeleconnection Indices To Forecast Short Term Drought

Class Transitions In Sicily (italy)

Brunella Bonaccorso1, Antonino Cancelliere2, Giuseppe Rossi2

bbonaccorso@unime.it

1 Department of Civil, Computer, Construction and Environmental Engineering and AppliedMathematics, University of Messina

2 Department of Civil Engineering and Architecture, University of Catania

An effective drought management can be pursued only if reliable monitoring andforecasting tools, able to promptly warn against the occurrence of forthcomingsevere events, are in operation. Among the several proposed drought monitor-ing indices, the Standardized Precipitation Index (SPI) has found widespreaduse to monitor dry and wet periods of precipitation aggregated at different timescales. Many recent studies have analyzed the role of SPI for drought forecast-ing, with particular regard to the assessment of transition probabilities betweenSPI drought classes. Furthermore, the influence exerted by some prominentpatterns of large scale atmospheric circulation variability (commonly referredas to “teleconnections”), such as El Niño Southern Oscillation (ENSO), NorthAtlantic Oscillation (NAO) or European Blocking (EB), on regional climatic fea-tures has been investigated and some effects of NAO on European climate hasbeen observed and analyzed in several studies.

In the paper two probabilistic models to forecast one, two and three-monthahead SPI drought class transition probabilities from a current SPI droughtclass and from a current SPI value, respectively, are presented and extended inorder to include information provided by an exogenous variable, such as a tele-connection index. Both models have been applied and tested with reference toSPI series computed on average areal precipitation in Sicily region (Italy), mak-ing use of NAO as exogenous variable. Results seem to indicate that droughttransition probabilities in Sicily are generally affected by NAO index. Further-more, application of a simple score approach to quantitatively assess the skill inforecasting of the proposed models, shows that the one which assesses transitionprobabilities from current SPI and NAO values to future SPI classes outperformsthe other models.

74

An impact perspective on pan-European droughtsensitivity

Kerstin Stahl1, Irene Kohn1, Lucia De Stefano2, Lena M. Tallaksen3, FranciscoM.C. Castro Rego4, Dionysis Assimacopoulos5, Antonio Massarutto6, Sonia I.

Seneviratne7, Henny H.A.J. van Lanen8

kerstin.stahl@hydrology.uni-freiburg.de

1 University of Freiburg ; 2 Complutense University of Madrid ; 3 University of Oslo;4 University of Lisbon ; 5 National Technical University of Athens ; 6 Bocconi University;

7 ETH Zürich ; 8 Wageningen University

In the past decades, Europe experienced several severe drought events with di-verse environmental and socio-economic impacts. Based on a number of differentdatasets and methods the EU FP-7 project DROUGHT R&SPI has investigatedthis diversity and the potential causes for similarities and differences in pastdrought impacts across different European countries and geoclimatic regions.The establishment of the European Drought Impact report Inventory (EDII)played a central role in this investigation. This database already contains over2500 entries and is now available to the public online allowing for more entriesfrom the community. All textual impact reports from reliable sources are clas-sified into 15 impact categories representing different sectors and a number ofsub-types. They are referenced in time and space to allow mapping and link-age to other information such as quantitative information on the hydroclimaticdrought hazard and on factors that influence regional to local drought vulnerabil-ity. The collection and analyses that were carried out over the project’s durationsuggest many commonalities across Europe, for instance the dominance of im-pacts on agriculture. Impacts on water supply, energy and industry, however,are less comparable across Europe reflecting rather different water use systemsand association with different drought types and specific vulnerabilities. Hence,for monitoring and prediction of impacts on these sectors has to be customizedregionally to locally. This contribution presents the similarities and differencesfound in the collection of impact reports across Europe and the lessons learnedfor the assessment of vulnerability and risk across diverse geoclimatic regions. Itfurther outlines suggestions for the continuation of the impact report collectionbeyond the project’s life span and potential of its further use.

75

BIODROUGHT project – developing effective tools forretrospective bioindication of dry episodes in streamhistory and selection of watersheds with drought risk

using aquatic macroinvertebrates

Petr Paril1,2, Světlana Zahrádková1,2, Marek Polášek1,2, Vít Syrovátka1,2,Michal Straka4, Lenka Šikulová2,4, Denisa Němejcová1, Pavla Řezníčková3

paril@sci.muni.cz

1 T. G. Masaryk Water Research Institute, public research institution, Mojmírovo náměstí16, 612 00

Climate change has brought remarkable changes in hydrological regime over thelast decades and causes consequent periodical drying out of water-courses (meantas the absence of surface flow) not only in the Mediterranean but also in CentralEurope where the percentage of annually drying up small streams has increasedremarkably. Increasing stream intermittency in this region is probably associatedwith changes in the distribution of precipitation and higher evapotranspirationduring the season, and also with hydrological alterations of anthropogenic origin.Regardless of the cause of drying up of small water-courses, we use the analysisof aquatic macroinvertebrate community as an effective tool for the detection ofdry periods in stream’s recent history.

The main expected project outcomes include (i) risk maps assessing streamvulnerability to drought and (ii) a bioindication method evaluating the presenceand possibly the extent of dry periods with a defined probability rate. Dueto lack of long term data for the evaluation of drought-related impacts on themacroinvertebrate community structure in the Central Europe, we analyzed datafrom the Czech state monitoring database SALAMANDER (collected in 1996-2010) and our own data collected within the ongoing BIODROUGHT project(2012-2015), www.biodrought.eu. We preclude that drought acts as an ecologi-cal filter which leaves its “footprint” in macroinvertebrate community structurefor a certain time period and the structural changes in species composition andrepresentation of species traits correspond with the spatial and temporal ex-tent of drought impairment. While the drought sensitive taxa (i.e. potentialpermanency indicators) are depleted, the proportion of advantageous traits and

76

drought resistant taxa or good colonizers increases and stays on higher level fora specific time period after stream re-flooding. We want to incorporate tradi-tional and frequently used indices (e.g. saprobic index) calculated on the basisof macroinvertebrate community composition in drought indication, but alsonew metrics are tested. This unique retrospective approach, which employs thecombination of the presented methods (indicator taxa analysis, species traitsanalysis and indices), can be used not only to uncover the previous dry episodes,but also to validate the developed drought risk maps.

Preliminary results, which are only available now, show lower abundancesand number of taxa in intermittent streams and remarkable differences betweenintermittent and permanent streams in terms of the representation of specifictraits (e.g. flow velocity preferences, locomotion type, voltinism etc.). Theseresults are very promising, although we must keep in mind that the response ofmacroinvertebrates may differ slightly in relation to local conditions (e.g. refu-gia availability, recolonisation sources or hydromorphological status of a watercourse). Regardless of certain variability in community response to drought, wedeveloped useful retrospective method which provides reliable results, unattain-able by conventional tools.

The project is supported by the Technology agency of the Czech Republicno. TA02020395.

77

Assessing future droughts economic losses with generichydroeconomic modelling

Noémie Neverre1, Patrice Dumas1

neverre@centre-cired.fr

1 CIRED (Centre International de Recherche sur l’Environnement et le Développement) /CNRS (Centre National de la Recherche Scientifique)

The aim is to evaluate the impacts of global changes on water scarcity in theMediterranean. We develop a generic hydroeconomic model able to confrontfuture water supply and demand on a large scale, taking into account man-madereservoirs. The assessment of the cost is done at the scale of river basins, usingonly information available at a regional scale. The methodology can thus begeneralized.

On the supply side, we evaluate the impacts of climate change on water re-sources. Reservoirs are located using Aquastat. Runoff is taken from the outputsof CNRM Meteo France climatic model (Dubois et al., 2010). Sub-basin flowaccumulation zone of each reservoir is determined based on a Digital ElevationModel (hydro1k). On the basis of this information the available quantity ofwater at each site is computed.

On the demand side, agricultural and domestic demands are projected interms of both quantity and economic value.

Globally available data on irrigated areas and crops are combined in orderto determine irrigated crops localization. Then, crops irrigation requirementsare computed for the different stages of the growing season using Allen (1998)method with Heargrave potential evapotranspiration. Irrigation water economicvalue, the opportunity benefit of water, is based on a yield comparison approachbetween rainfed and irrigated crops. Potential irrigated and rainfed yields aretaken from LPJmL (Blondeau et al., 2007), or from FAOSTAT by making simpleassumptions on yield ratios.

For the domestic sector, we project the combined effects of economic devel-opment and water costs evolution on future demands. The method consists inbuilding 3-blocks inverse demand functions where volume limits evolve with thelevel of GDP per capita, with a saturation (similar to WaterGAP methodology).

78

The value of water along the demand curve is determined from elasticity, priceand quantity data from the literature, using the point-extension method, andfrom water costs data.

Then projected demands are confronted to future water availability. Operat-ing rules for water allocation between demands are based on the maximizationof water benefits; a parameterisation-simulation-optimisation approach is used.This gives a projection of future droughts in the different locations and an esti-mation of the associated direct economic losses from unsatisfied demands.

The regional generic hydroeconomic model will be applied to North Africa.

79

Low flow and drought in Europe - lessons from the past

Lena M. Tallaksen1

l.m.tallaksen@geo.uio.no

1 Department of Geosciences, University of Oslo, Norway

Low flow and drought studies in Europe have gained increased attention in recentdecades and achieved recognition as an important research area. Notable, theEuropean FRIEND project put the topic on the international research agendain the 1980’s and contributed with several reports and scientific papers on lowflows in particular. A special focus was given to the derivation of low flowstatistics, regional analyses (beyond the national scale) and estimation at theungauged site. With the EU-project ARIDE (Assessment of the regional im-pact of droughts in Europe) covering the period 1998-2000, focus shifted partlyfrom low flows to drought, with a motivation to increase knowledge on the de-velopment, extent and spatial characteristics of drought at the European scale.ARIDE was followed by a series of EU projects on drought, which with thesupport of numerous nationally studies, significantly have increased our under-standing of drought as a natural hazard, its dynamic nature and generating pro-cesses across the wide range of hydroclimatic regions in Europe. Studies havenot been limited to hydrological drought, but have explored different types ofdroughts, including the propagation from its origin as a meteorological droughtthrough reduced soil moisture and finally, manifestation as a drought in stream-flow and groundwater. Advances have been made in the identification of majorpan-European events, their characteristics and drivers as well in the effort toidentify impact-relevant indicators for drought monitoring. However, challengesstill remain in reproducing low flow dynamics by hydrological (and land surface)models, which influence the ability of such models to satisfactory map spatialand temporal patterns of large-scale hydrological droughts. A joint outcome ofseveral drought projects has been the establishment and continuous developmentof the European Drought Center1 (EDC), a community effort across institutionsand projects. The EDC recently became host for the European Drought Ref-erence (EDR) database and the European Drought Impact Report Inventory(EDII) developed as part of the EU project DROUGHT-R&SPI2. A better un-

80

derstanding of the links between physical indicators and key drivers of drought isvital for the application of this knowledge to drought forecasting, whereas a bet-ter understanding of the links between physical indicators and drought impactsis critical to improve drought preparedness and support drought managementplans. This presentation highlights key achievements in low flow and droughtresearch in Europe with a special emphasis on the identification of low flow anddrought episodes or events, detection of recent changes, ability to model low flowbehavior and key characteristics and main drivers of major historical events.

81

Droughts and low flows in present and future climate –pacing the path between climatology and hydrology

Klaus Haslinger1

klaus.haslinger@zamg.ac.at

1 Climate Research Dep., Central Inst. for Meteor. and Geodynamics (ZAMG), Austria

Climate change impacts on low flows and droughts may have a significant im-pact on socio-economic systems. In this talk selected results of the Austrianproject initiative CILFAD (Climate Impacts on Low Flows And Droughts) arepresented, comprising the following topics: (i) linking meteorological drought in-dices to streamflow, (ii) assessing future drought probabilities based on regionalclimate model simulations and (iii) merging multiple methodological approachesfor robust future low flow and drought projections.

The first topic deals with the interplay between atmospheric forcing andcatchment response on low flows and streamflow anomalies. A comprehensiveanalysis between meteorological drought indices and stream flow indices wascarried out to quantify the strength of the relationship. The focus of that analysiswas to decipher major influences on the characteristics of these links and todistinguish between climatic and catchment drivers.

The second topic deals with atmospheric projections of drought characteris-tics in Austria and the Greater Alpine Region. Regional Climate Model (RCM)simulations under three different emission scenarios were used to quantify chang-ing probabilities in future drought occurrence and also to assess spatial patternsand main drivers (precipitation or evapotranspiration) for drought probabilitychanges.

In the last part, both types of information are combined into a synopticview of future low flows and droughts. This includes assessments of streamflowtrends as output of the system, but also rainfall-runoff projections of atmosphericprojections, and stochastic projections which combine gained knowledge aboutatmospheric drought signals and the link to streamflow drought. The synopsisof different approaches rooted between the fields of climatology and hydrology,shall enable more consistent estimates of past, present and future drought.

82

National Drought Mitigation Center Drought Services:Collaborations Towards a Global Drought Early Warning

and Information System

Mark D. Svoboda1

msvoboda2@unl.edu

1 National Drought Mitigation Center University of Nebraska-Lincoln, USA

Drought is a hazard that lends itself as well to diligent monitoring and earlywarning. Unlike most hazards, the fact that droughts typically evolve slowly,can last for months or years and cover vast areas spanning multiple politicalboundaries/jurisdictions and economic sectors can make it a daunting task tomonitor, mitigate and plan for.

The National Drought Mitigation Center (NDMC) works to reduce societalvulnerability to drought by helping decision makers at all levels to: implementdrought early warning and decision support systems, understand and preventdrought impacts and increase long-term resilience to drought through proactiveplanning. The NDMC is a national center founded in 1995 at the University ofNebraska-Lincoln in the United States. The NDMC conducts basic and appliedresearch along with the development and maintenance of a number of operationaldrought-related activities, including the U.S. Drought Monitor (USDM), NorthAmerican Drought Monitor (NADM), Drought Impact Reporter (DIR), DroughtRisk Atlas (DRA) and the Vegetation Drought Response Index (VegDRI).

In addition to our coordinated efforts with the National Integrated DroughtInformation System (NIDIS) (drought.gov), the NDMC is working with manypartners around the world to help form a collaboration and coordination nexuswith an ultimate goal of building comprehensive global drought early warningand information systems (GDEWIS).

This presentation will describe in more detail the various drought resources,tools, services and collaborations already being provided by the NDMC and itspartners along with a look at what is in the works for the future in helping otherstoward developing drought early warning and information systems in the U.S.and around the world.

83

Multisectorial Partnerships for drought management andmitigation. The experience of the Jucar River Basin

(Spain)

David Haro1, Abel Solera1, Javier Paredes-Arquiola1, Joaquín Andreu1

David.Haro.Monteagudo@gmail.com

1 Institute of Water and Environmental Engineering, Universitat Politècnica de València

The Jucar River Basin District (JRBD) is one of the most vulnerable areasto drought of the western Mediterranean region due to climate characteristics(mostly semiarid), high water exploitation indexes (water scarcity), and veryhigh space and time variability of precipitation (leading to a highly seasonal, andinter-annual variability in the river flows). Hydrological droughts in this regionare frequent, and they can be very intense and with duration of years (frequentlymore than 3 years). For these reasons, adaptation to drought has been a constantobjective of people living in this area; especially water users (mostly irrigateddistricts and cities). Adaptation has been carried out by infrastructure develop-ment, but also by organizational innovation. Therefore, partnerships have beendeveloped since many years ago in the area.

Initially, single-sectorial partnerships were predominant, but in 1936 an MSPwas created which included all major sectors of water uses, the main bodiesof the central administration related to water, as well as provincial and localrepresentatives. Its name in Spanish (Confederación Hidrográfica del Júcar -CHJ) literally means Júcar River Basin Partnership. Later, the diversificationof interests within the CHJ related to the different aspects of water planningand management revealed the necessity of dividing the decision making processaspects into several internal bodies. These bodies have always had the intention-ality of including all, or the most, of the voices interested in the topics addressed.Therefore, a cluster of satellite MSPs have been created along the years to dealwith the different problems existing within the JRBD. In this document, we willdescribe the existing, or envisaged, MSPs related to drought risk managementand mitigation to a higher or a lower extent. Among them, we will focus on thePermanent Drought Commission (PDC), including representatives of the samepartners (sectors and stakeholders) mentioned before, and which is in action in

84

times of drought, until recovery to normality. The importance of the PDC isvery high, since its decision will influence the management and mitigation of thedrought episodes, when very high potential damages and risk to economy andhuman safety can happen.

In the present document, the genesis and objectives of the MSPs will be de-scribed, as well as the stakeholders and their role. Besides of the CHJ and thePDC, other satellite and collateral MSPs will be mentioned and depicted, mainlyin order to show how different MSPs can interact and cooperate for the generalobjective of drought hazard risk reduction. The Special Plan for Drought Plan-ning and Management designed by CHJ will also be mentioned and explained,since it is nowadays the basis for a proactive approach against droughts. Mea-sures available to the MSPs to reduce risk and to improve resilience will bedescribed, including economic instruments. And also, the transferability of theexperiences to other regions and countries will be explored.

85

Towards a comprehensive drought information system forEurope

Jürgen Vogt1, A. de Jager1, D. Magni1, L. Valentini1, F. Micale1, C.Cammalleri1, G. Sepulcre1, C. Lavaysse1, S. Russo1, P. Barbosa1, J. Spinoni1

juergen.vogt@jrc.ec.europa.eu

1 European Commission, Joint Research Centre, Institute for Environment andSustainability (JRC-IES), Ispra, Italy

Europe has repeatedly been affected by droughts, resulting in considerable eco-logical and economic damage and climate change studies indicate a trend towardsincreasing climate variability most likely resulting in more frequent drought oc-currences also in Europe.

Against this background, the European Commission’s Joint Research Centre(JRC) is developing methods and tools for assessing, monitoring and forecastingdroughts in Europe and develops a European Drought Observatory (EDO) tocomplement and integrate national activities with a European view.

At the core of EDO is a portal, including a map viewer, a metadata cata-logue, a media-monitor and analysis tools. Underlying data stem from groundand satellite observations as well as from distributed hydrological models andare stored in a relational database. Through the map viewer Europe-wide up-to-date information on the occurrence and severity of droughts is presented,complemented by more detailed information from regional, national and localobservatories through OGC compliant web-mapping services. The continent-wide meteorological, soil moisture-related and vegetation-related indicators arethen integrated into combined indicators showing different alert levels targetedspecifically to decision makers in water and land management. Finally, timeseries of historical maps as well as graphs of the temporal evolution of droughtindices can be retrieved and analysed.

On-going work is focusing on developing reliable medium and long-rangeprobabilistic drought forecasts, the assessment of drought impacts in differentsectors, the analysis of climate change impacts on drought occurrence, durationand severity as well as the analysis of current and future drought hazard andrisk.

86

As a tool EDO contributes to the development of a Global Drought Infor-mation System (GDIS) and provides support to the Integrated Drought Man-agement Programme (IDMP). With its multi-level setup EDO could serve asa model on how to build a global drought information system from droughtobservatories on different continents and at different spatial scales. In this con-text similar set-ups for drought observatories for Africa and South and CentralAmerica have been tested in the frame of research and collaboration projects.

The presentation will provide an overview on the development and state ofEDO, the different products, as well as options for further development.

87

Drought Duration Curves: A method to quantifycontinental differences in hydrological droughts

Erik Tijdeman1, Kerstin Stahl1, Sophie Bachmair1, Jamie Hannaford2

erik.tijdeman@hydro.uni-freiburg.de1 University of Freiburg

2 Centre for Ecology & Hydrology

Droughts are globally occurring natural hazards which have been shown to havesevere impacts in a variety of sectors and on a variety of scales. Monitoring andEarly Warning (M&EW) systems help society to be better prepared for upcom-ing drought events. Typically, these systems are based on drought indicatorswhich have rarely been tested against observed drought impacts. The DRIVERproject is seeking to compare available drought indicators with drought impacts,using some of the most advanced available M&EW systems on a continental scale(which can be found in the United States, Australia and Europe). However, thisanalysis of indicator-to-impact relationships requires detailed knowledge aboutdifferences in drought characteristics on these continents, in order to providea “baseline” of droughts in these widely-varying climatic and geographical set-tings. This is particularly important for hydrological (i.e. streamflow) droughtsthat are generally one of the more under-developed parts of M&EW systems andwhich, compared to meteorological or agricultural droughts, are more heavily in-fluenced by landscape properties such as catchment topography, soils, geology,etc.

The aim of this study is therefore to compare drought characteristics in near-natural streams on these three continents, using a dataset of around 1000 catch-ments. Drought Duration Curves (DDC), a cumulative distribution of droughtdurations, are computed per station and used as a proxy for this comparison.Ensembles of these DDC are grouped per continent and their properties areutilized to quantify differences in drought characteristics.

A second step is to explain continental differences in drought characteristics.What are the main drivers for these differences? And are these drivers mainlybased on climate or physical catchment properties? To answer these questions,different subsets of DDC ensembles are selected using grouping criteria such asclimate, precipitation, average basin elevation or soil type.

88

The results show continental differences in drought characteristics and theinfluence of different drivers on these characteristics. Furthermore, these resultsare used to see whether similar drivers (e.g. soil type X or climate Y) show similarcharacteristics on different continents. The homogeneity within the differentsubsets of DDC ensembles serves as a proxy to test the significance of grouping.

The methodology was found to provide useful information. Nevertheless,the drought identification method applied in this study is only one of the manydrought identifications methods available. Robustness of this analysis needs tobe obtained by testing different drought identification methods. In the end,further work will relate the different hydrological drought indicators with eachother, with the ultimate goal to find the indicator(s) that can best be linked toreported drought impacts across these diverse study areas.

89

The influence of catchment parameters on droughtoccurrence in the Tatra Mts., Slovakia

Miriam Fendekova1, Marian Fendek1

fendekova@fns.uniba.sk

1 Comenius University in Bratislava

Groundwater runoff and its changes were evaluated for 57 catchments of theSlovak part of the Tatra Mts. Massif. The Tatra Mts. are located in the north-ern part of Slovakia and create a natural border with the Poland. They can bedivided into three parts: (1) western, called Zapadne Tatry Mts. or Rohace, (2)central, called Vysoke Tatry Mts., and (3) eastern part, called Belianske TatryMts. These three parts differ in geological structure and rocks they are built of,as well as by climatic conditions. Western part is built of metamorphic rockswith the Mesozoic carbonatic envelope; central part is built of granitic rockswith very spare occurrence of Mesozoic carbonates. Mesozoic rocks of Trias-sic age prevail in the eastern part. The western part is the most wind-wardlocated part, the central and eastern part are precipitation-shaded in a differ-ent rate. Time series longer than 30 years were at the disposal at 16 dischargegauging profiles. The rest of time series had the shorter length; most of themwere observed within the hydrogeological investigation works in the past. Allthe data was statistically processed. Daily base flow values were derived fromthe discharges using the local minimum method, long-term average values werecalculated using the methods of Kille and Castany. Base flow values were re-calculated into specific base flows in order to be able to compare the differentcatchments. Specific groundwater runoff values in geologically different catch-ments were inter-compared within respective hydrological decades. The resultsshowed differences in respect to groundwater runoff values and their seasonality.After that surface and groundwater drought occurrence was analyzed using theThreshold Level and the Sequent Peak Algorithm methods. Research resultswere compared with known data on groundwater drought occurrence in otherareas of Slovakia and some other countries.

90

An assessment of actual evapotranspiration and soil waterdeficit in agricultural regions in Europe

Blaz Kurnik1

blaz.kurnik@eea.europa.eu

1 European Environment Agency

Changes in agricultural droughts were investigated using simulations of soil wa-ter deficit (SWD) and actual evapotranspiration (ETA) from a distributed semi-empirical soil water balance model - swbEWA. At the European scale both SWDand ETA did not change significantly between 1951 and 2011. However, signifi-cant increases in SWD were found in southern Europe, except in western Turkey,whereas in northern Europe changes in SWD remain small. ETA increased sig-nificantly as a joint response to increased air temperature and precipitation innorthern Europe.

Empirical Orthogonal Function (EOF) analysis applied to the model outputsdemonstrate that large-scale agricultural droughts are influenced by the recur-rence of the North Atlantic Oscillation (NAO) and by atmospheric blocking.This has been achieved by calculating Pearson correlation coefficients (RPear-son) between SWD and NAO and atmospheric blocking for each month of theyear.

During a negative phase of NAO, storms over the North Atlantic and Europeare less frequent and as a consequence dry weather in Europe is observed. Pos-itive NAO influences agricultural drought in Europe by shifting storms tracksfrom the North Atlantic onto European continent to more northerly paths, whichin turn decreases the amount of precipitation over central Europe.

Large soil water deficits are mainly influenced by atmospheric blocking. No-tably, winter blocking increases the severity of agricultural droughts in south-western Europe, while summer blocking influences agricultural droughts in south-eastern Europe. Notwithstanding, the first three EOFs contribute to less than40 % of the total spatial variability of soil water deficit. This shows that agri-cultural droughts are complex phenomena that can be only partly explained byextreme NAO or by intensive atmospheric blocking.

91

Response of simple stream hydro-chemical indicators torainfall-runoff events in forest disturbed basins in upper

Vydra, Central Sumava Mountains

Ye Su1, Jakub Langhammer1, Jana Kaiglova2, Jerker Jarsjö3

leafsuning@gmail.com

1 Faculty of Science, Charles University in Prague, Albertov 6, 128 43 Praha 2, CzechRepublic

2 DHI Hydroinform a.s., Na Vrších 5, 100 00, Praha 10, Czech Republic Faculty of Science,Charles University in Prague, Albertov 6, 128 43 Praha 2, Czech Republic

3 Department of Physical geography and Quaternary Geology, Stockholm University, 106 91,Stockholm, Sweden

In montane regions, the quantity and quality of forested basins streamflow iscomplex due to spatial and temporal variability of the hydrological processesand land cover. Simple water quality parameters can be used as indicators tounderstand the contributions of sub-streams, in-stream hydrological behaviors,and runoff formation.

We therefore aim to i) test applicability of the monitoring from automatedsensor networks, coupling high frequency water level measurement and detectionof simple hydro-chemical parameters, namely electronic conductivity (EC), foranalysis of runoff response of montane catchments; ii) to use Q-EC patterns toidentify catchment’s response to different types of rainfall-runoff events; and iii)to use Q-EC patterns to distinguish differences in runoff response of catchmentswith different extents and types of forest disturbance or damage, helping tobetter understand the changing dynamics of runoff processes in the disturbedareas.

Six experimental catchments within the upper Vydra basin, in the CentralSumava Mountains, were chosen as the study sites. Each of these six catch-ments has an area approx. 2-5 km2 and covers different types of physiographicconditions, land cover related to forest properties and disturbance. The au-tomatic monitoring network of the experimental catchments consists of waterstages, WT, and EC measurements at the outlets, among which 3 stations havecontinuously collected meteorological data (precipitation, snow depth, and airtemperature) in 10-minute interval since 2009.

92

In the study, from 73 events occurred in 2008-2013, different rainfall-runoffevents–storm/rain on dry/wet condition, and long-term rain were classified to6 groups by using Agglomerative hierarchical clustering (AHC) method. Ingeneral, the EC performed in a positive correlation with Q. Specifically, eachtype of events result in significantly different responses of Q and EC level andthe Q-EC loop pattern (shape, slope, and direction). Moreover, the responsesof Q-EC significantly differ among catchments with moderate forest disturbance(PTA &CIK), heavy forest disturbance (BRE) and clear-cutting (ANT). Theseresponses to the rainfall-runoff events, the loop of each hydro-chemical parametergive information about the in-stream flow pathways in different catchments withvarying forest cover.

The contribution of this study gives an alternative to high-cost techniquesand an application of using simple natural tracers to indirectly quantify rainfall-runoff processes in stream as well as the role of sub-stream to the whole basin.

93

Identifying a global drought structure using the copulamethodology

Franziska Gaupp1

franziska.gaupp@ouce.ox.ac.uk

1 Oxford University

With a focus on global food security, my study will look at drought incidentsin different regions and investigate globally correlated drought events and theiroverall influence on agriculture. Looking at past droughts in different regions ofthe world in a short time-span and their joint effects such as price spikes on theworld grain market show clearly that it is important to understand correlatedextreme weather events. Especially under climate change, there is a need toinvestigate the probabilities of correlated droughts and their influence of agri-cultural production.

This study focuses on droughts and drought modelling. For each major cropproducing area (“breadbasket”) of the world the drought pattern will be inves-tigated. Drought characteristics such as duration, frequency or severity will beexamined in each breadbasket separately and the joint probability distributionof several characteristics will be calculated. Then, a global dependency struc-ture of droughts will be identified by linking the probability distribution of acertain drought characteristic in one breadbasket with the distribution of thesame variable in another breadbasket using the copula methodology. In thatway, the probability of a severe drought in different breadbaskets at the sametime can be estimated.

In order to identify drought, theWeighted-Anomaly-of-Standardized-Precipitation(WASP) index by Lion and Barnston (2005) and the Standardized Runoff In-dex (SRI) are used. Furthermore, the correlation between large-scale circulationindices such as the North Atlantic Oscillation (NAO) or the El Niño-SouthernOscillation (ENSO) and droughts (Dai et al., 2004; Hannaford et al., 2011) areused to identify a global dependence structure of droughts.

Due to their multidimensionality, droughts require joint modelling of differentrandom variables such as drought frequency, duration and severity. In this study

94

the copula methodology is used to capture the different aspects of droughts. Cop-ulas are functions which combine univariate distribution functions in order toform multivariate distribution functions (Shiau, 2006; Shiau et al., 2007). Thecopula method is a flexible tool as it allows marginal distributions from differentfamilies to describe and to model dependence between random variables. Espe-cially in drought modelling, the copula method yields clear advantages over asimple bivariate probability distribution as droughts are complex, multivariatehydrological events characterized by several correlated random variables (Shiau,2006). Drought duration is usually exponentially distributed and drought sever-ity is gamma distributed (Shiau, 2006; Shiau et al., 2007; Khedun et al., 2011)which makes it impossible to use conventional bivariate distributions as theyrequire the same family for each marginal distribution.

This paper will present an in-depth drought analysis in the major wheat,maize and soybean producing regions in the US, Brazil, France, Russia, Indiaand China investigating the correlation between drought frequency, durationand severity. Furthermore, results will be presented that show a dependencestructure between droughts in different breadbaskets indicating risks for globalcrop production.

References:Dai, A., Trenberth, K. E., Qian, T. (2004) A global dataset of Palmer Drought

Severity Index for 1870–2002: relationship with soil moisture and effects of surfacewarming. Journal of Hydrometeorology, 5(6), pp. 1117-1130.

Hannaford, J., Lloyd-Hughes, B., Keef, C., Parry, S., Prudhomme, C. (2011) Ex-amining the large-scale spatial coherence of European drought using regional indicatorsof precipitation and streamflow deficit. Hydrological Processes, 25(7), pp. 1146-1162.

Khedun, C. P., Chowdhary, H., Giardino, J. R., Mishra, A. K., Singh, V. P. (2011)Analysis of Drought Severity and Duration Based on Runoff Derived from the NLSM.

Lyon, B., Barnston, A. G. (2005) ENSO and the spatial extent of interannualprecipitation extremes in tropical land areas. Journal of Climate, 18(23).

Shiau, J. T. (2006) Fitting drought duration and severity with two-dimensionalcopulas. Water Resources Management, 20(5), pp. 795-815.

Shiau, J. T., Feng, S., Nadarajah, S. (2007) Assessment of hydrological droughtsfor the Yellow River, China, using copulas. Hydrological Processes, 21(16), pp. 2157-2163.

95

Investigation of Drought Occurrence and Propagationbetween Meteorological, Agricultural, and Hydrological

Drought

Ji Yae Shin1, Jiyoung Yoo2, Minsung Kwon3, Tae-Woong Kim1

starzee@hanyang.ac.kr

1 Hanyang University, South Korea2 Chonbuk National University

3 Sungkyunkwan University

This study investigated the relationship between meteorological, agricultural,and hydrological drought in South Korea, from perspectives of their occurrenceand propagation. The primary cause of a drought is the lack of precipitation overa large area and for an extensive period of time. Considering various perspec-tives and interests of drought, there are three main types of drought commonlyused in practice. Meteorological drought is usually defined based on the de-gree of dryness and the duration of the dry period, which is a departure fromthe normal hydro-climatic features of a region. Agricultural drought focuseson the drought impact on agriculture like soil water deficits, reduced ground-water or reservoir levels needed for irrigation. Hydrological drought usuallyoccurs following periods of extended precipitation shortfalls that affect watersupply, potentially resulting in significant societal impacts. Since most regionsare closely interconnected by hydrologic systems, the impact of deficit of wa-ter/precipitation is propagated through the hydrological cycle. In this study,employing various drought indices to represent meteorological (e.g., Standard-ized Precipitation Index (SPI)), agricultural (e.g., Crop Moisture Index (CMI)),hydrological drought (e.g. Palmer Hydrological Drought Index (PHDI)), thecharacteristics of occurrence and propagation of drought are evaluated. Therelationship between meteorological, agricultural, hydrological drought is alsoinvestigated through calculation of class transition probabilities and graphicalrepresentation with actual drought response actions such as enforcing restrictivewater rationing for specific regions.

96

A drought index accounting for snow

Maria Staudinger1, Jan Seibert1, Kerstin Stahl2

maria.staudinger@geo.uzh.ch

1 Department of Geography, University of Zurich2 Chair of Hydrology, Albert-Ludwigs University Freiburg

The Standardized Precipitation Index (SPI) is the most widely used index tocharacterize droughts that are related to precipitation deficiencies or long pre-ceding periods of precipitation shortages. However, the SPI does not alwaysdeliver the relevant information for hydrological drought management particu-larly in snow influenced catchments. If precipitation is temporarily stored assnow, then there is a significant difference between meteorological and hydro-logical drought because the timing of the melt and rain water being releasedto the stream differs from the timing expected from the SPI. We introduce anextension to the SPI, the Standardized Snow Melt and Rain Index (SMRI), thataccounts for rain and snow melt deficits, which effectively influence streamflow.The SMRI does not require snow data, but is based

solely on temperature and precipitation. The value of the new index is illus-trated for seven Swiss catchments with different degrees of snow influence. Forthese catchments the SMRI was found to be an important complementary indexto the SPI to characterize hydrological droughts.

97

Identification of Consecutive Dry Days (CDD) usingobservational and model data from COSMO CLM

Livia Labudova1, Lukas Schefczyk2, Milan Trizna1

livia.labudova@gmail.com

1 Department of Physical Geography and Geoecology, Faculty of Natural Sciences, ComeniusUniversity in Bratislava Mlynska dolina, 84215 Bratislava

2 Department of Environmental Meteorology, Faculty of Regional and EnvironmentalSciences, University of Trier, Behringstr. 21, 54296 Trier

Several drought indices have been developed to identify drought periods in recentdecades. In this paper, we report on drought events occurrence analysis using theworldwide established Consecutive Dry Days (CDD) index. This index, basedonly on daily precipitation,is on the list of 27 indices, which are recommended forclimate change monitoring by the Expert Team on Climate Change Detectionand Indices (ETCCDI). Two water basins in Slovakia were selected as studyareas according to their different precipitation regimes and the impact of theNorth Atlantic Oscillation (NAO). For the drought events occurrence analysis,observational data in the period 1981 – 2010 and model data in the period 1981- 2000 were used.The model data origin from the COSMO CLM (COnsortium ofSMall scale MOdelling – in CLimate Modus) model with a horizontal resolutionof 18 km. The analysis is focused on the seasonal drought occurrence and theirtrend in last decades. It is supplemented by precipitation trend analysis and thecomparison of the two datasets. This comparison shows that the model datatend to be wetter than the observations in the northern Slovakia (the Kysucawater basin), where annual precipitation is higher than in the southern Slovakia(the Slana water basin). In the latter, both datasets are in quite good agreement.

Acknowledgments: This study was the part of a DBU doctoral research grantsupported also by research grant UK/119/2014.

98

Spatio-temporal characteristics of wet and droughtextremes over territory of Russia

Dilyara Utkuzova1

trini_di@mail.ru

1 Hydrometeorological centre of Russia

Synoptical-statistical analysis of drought and wetness extremes for the territoryof Russia has been conducted using SPI index calculated for 500 stations withrecords from 1966 through 2010. Different parameters of SPI frequency distribu-tion and long-term tendencies were calculated as well as spatial characteristics in-dicating drought and wetness propagation. Results of analysis demonstrate thatduring last years there is a tendency of increasing of the intensity of draught andwetness extremes over Russia. There are fewer droughts in the northern regions.The drought propagation for the European territory of Russia is decreasing inJune and August, and increasing in July. The situation is opposite for the wet-ness tendencies. For the Asian territory of Russia, the drought propagation issignificantly increasing in July along with decreasing wetness trend.

99

Assessing the drought risk based on the joint probabilityof precipitation and soil moisture in Rhineland-Palatinate,

southwest Germany

Zhiyong Liu1, Lucas Menzel1

zhiyong.liu@geog.uni-heidelberg.de

1 Institute of Geography, Professorship in Hydrology and Climatology, Heidelberg University,Heidelberg, Germany.

A bivariate probabilistic model is presented in this study to model the depen-dence between precipitation and soil moisture. Conditional drought risks underdifferent precipitation scenarios are evaluated based on the established jointdependence structure. The methodology is illustrated using a case study inRhineland-Palatinate, southwest Germany. The gridded monthly precipitationand soil moisture data covering the period 1960-2008 are extracted from a hy-drological modeling system (TRAIN) implemented at our institute, at a spatialresolution of 1 km x 1 km. This study defines the soil moisture as the droughtindex. First, a set of theoretical probability distributions are assessed to find themost-fitted marginal distribution to precipitation and soil moisture. The jointdependence structure between precipitation and soil moisture is then establishedusing several different copula-based models. A Bayesian-integrated copula as-sessment method is used to assess the performance of the copulas and determinethe most appropriate copula. After establishing the joint dependence, the esti-mation of drought probability for a defined soil moisture threshold conditionedupon different precipitation scenarios becomes available. We assess spatial pat-terns of conditional drought probability (risk) under given precipitation condi-tions across Rhineland-Palatinate for each month. The pixels (locations) proneto occurring drought (high risk areas) and those sensitive to precipitation con-ditions are identified. The patterns of drought risk for different months underdifferent precipitation conditions across this study region are also characterized.These findings indicate that the copula-based approach can provide an addi-tional drought risk-estimation and thus would be beneficial for decision makers.The proposed methodology could also be promising to apply in other Europeanregions for drought risk estimation.

100

Stochastic modeling of spatio-temporal characteristics ofdrought in southern Iran

Arash Malekian1

malekian@ut.ac.ir

1 University of Tehran

Global warming, greenhouse effect, climate change, water resources depletion,and increased demand for more agricultural products put extra pressure on manyregions in the world. This is required to plan in the region in such a way as topreserve at least the present situation and, even better, to improve local con-ditions with the purpose of combating droughts and desertification. Efficientmanagement strategies should include rules and regulations to reduce the ef-fects of droughts by preparing for potential disaster maps. Temporal and spatialdrought characteristics come into the picture for a drought characteristics anal-ysis. Their predictions require different models including statistics, probabilitytheory, and stochastic processes. One of widely used methodologies for temporaldrought assessment is the standardized Precipitation Index (SPI), which is usedto quantify the rainfall deficit for several time scales. In the spatial droughtevaluation, drought occurrences at many sites are considered, and with the in-formation and knowledge at individual sites, regional drought characteristicsare evaluated. Spatial treatment of the droughts is mostly achieved by spatialanalysis techniques such as geostatistics. Meanwhile, drought has a stochasticnature and therefore, the future states of any stochastic process at a particu-lar location, cannot be predicted with certainty. However, based on past data,it is possible to assess the probability of any particular outcome. Stochasticprocesses in which the probability of a future state is dependent only on thepresent state and not on any of the past states are said to follow a first orderMarkov chains. Markov analyses study dependent random events—events whoselikelihood depends on what happened last, or where the sequence of occurrencemust be considered for analysis. Spatio-temporal characteristics of droughts forsouthern part of Iran using geostatistical and Markov chain approaches wereanalyzed in this research. A first order Markov Chain approach was used inthis research to deal with uncertain nature of drought and the matrix of tran-

101

sition probabilities was calculated based on SPI method. The SPI is calculatedfrom precipitation records, which after fitting to gamma distribution. Then, thecorresponding return period of droughts classes obtained for each station wascalculated and spatial continuity was obtained based on the resulted variogramsin each case. Contouring maps requires the use of some type of interpolationprocedure. Geostatistical approach was used at this stage to prepare iso- proba-bility and iso- return period maps of drought in the region. The calculated mapsshow that susceptibility of the region to droughts and water shortage which is avery useful tool for planners and decision makers to prioritize the region basedon its susceptibility to drought as well as to implement relevant measures inorder to mitigate the effects of droughts.

102

Influence of the Water Channel Network Facility in theKiso River System

Yoshinobu Sato1

satogon@gmail.com

1 Faculty of Agriculture, Ehime University

The Japanese Islands are surrounded by ocean and having relatively large precip-itation caused by typhoon, frontal rain and winter snowfall. However, comparedwith continental large rivers, rivers in Japan are very short and steep and mostwater supplied in the river basin by precipitation is flowing down into the seaquite rapidly. Therefore, available water resources in Japan is limited and fre-quently suffered by water shortage.

In the present study, we focus on the Kiso River System (Kiso Sansen), oneof the largest river system in Japan (9,100 km2), consisting of three big riversystem (the Kiso River, Nagara River and Ibi River Basin). The river water inthe Kiso Sansen have been used as the source of agricultural, drinking, industrialand power generation. However, the region frequently suffers drought due to lowamount of rainfall in recent years; there have been restrictions on water intakefrom rivers on 15 times during the period from 1996 to 2005.

Under these circumstances, the project of water channel network facility inthe Kiso River System is now under reviewing. The water channel networkfacility in the Kiso River System will lead the water made available from theTokuyama dam (the largest dam in Japan) at the upper reaches of the Ibi Riverto the Kiso River, in order to provide municipal water for Nagoya city (popula-tion: approx. 2,2 million) and other cities in Aichi prefecture, and to improvethe river environment during the extreme drought event. This water channelwill provide 1,7 m3.s−1 of water to the Nagoya city and 2,3 m3.s−1 of water tothe Aichi prefecture and provide 12,0 m3.s−1 of water additionally for improvingriver environment.

In the present study, a drought analysis of Kiso River System using an Inte-grated distributed hydrological model (Hydro-BEAM) which can consider actualreservoir operations, water intake from river channels and the water transfer bythe virtual water channel network facility was conducted. The results showed

103

that the water channel network can mitigate the extreme drought and shorteningthe period of drought approx. 7 days per year).

Such information will contribute for the better understanding of the futuredisaster predictions and integrated water resources (low water) management.

104

Rainfall-runoff projections of low flows in Austria

Juraj Parajka1, Alberto Viglione1, Günter Blöschl1

parajka@hydro.tuwien.ac.at

1 TU Vienna, Austria

The main objective of this study is to investigate potential impacts of climatechange on low flows and droughts in Austria. The main idea is to simulate andcompare low flow indices and their changes by using two different approaches. Inthe first, a conceptual rainfall-runoff model (TUWmodel, Parajka and Viglione,2012) is calibrated in 338 catchments that cover a wide range of physiographicconditions in Austria. In order to assess the uncertainty of rainfall-runoff pro-jections, different variants of model calibration are performed. The model iscalibrated separately in three different decades (1976-86, 1987-97, 1998-08) indi-cating the impact of time stability of model parameters (Merz et al., 2011) andby using 11 variants of compound objective function (Merz et al., 2011), indi-cating the modeling uncertainty. The low flow projections are then estimatedfor all calibration variants and four future climate scenarios (ECHAM5-A1, A2,B1 and HADCM3-A1) by using a delta change approach. The final results areregionalised for the entire stream network of Austria by the Top-kriging method(Skøien et al., 2014).

In the second approach, a different strategy focusing on mechanisms of cli-mate induced low flow changes is applied. The mechanisms include increases inair temperature and hence increases in evaporation and changes in the timing ofsnowmelt and changes in seasonal precipitation and hence changes in soil mois-ture and low flows. The changes in air temperature and precipitation are derivedfrom the observed historical trends and their extrapolation at selected climatestations. Based on this strategy, a stochastic model for low flows is set up. Thissetup includes the calibration of stochastic rainfall model and the simulation ofan ensemble of 100 possible time series of rainfall and air temperature. Thisensemble is then used as an input for the TUWmodel simulations of low flowindices.

The rainfall-runoff projections indicate an increase of low flows in the Alps,typically in the range of 10-30 % and a decrease in south-eastern part of Austria

105

(Styria) mostly in the range of -5 - -20 % for the period 2021-50. The changein the seasonality varies between scenarios, but there is a tendency to earlierlow flows in the Northern Alps and shift to later occurrence in the EasternAustria. The uncertainty range from four climate scenarios is within 15-20 %at the majority of Austria, while the total model uncertainty is much higher,i.e. more than 50 % particularly in the Alps. This suggests the importance ofselecting objective functions for modeling low flows projections.

The stochastic projections of low flows indicate the important role of airtemperature changes. In some regions, the projections indicate a decrease in lowflows, even if annual precipitation does not decrease. In the Alps, a decreasein precipitation is compensating by increasing snowmelt, which indicates nosignificant change of low flow indices in the future.

References:Merz, R., J. Parajka, and G. Blöschl (2011) Time stability of catchment model pa-

rameters: Implications for climate impact analyses, Water Resour. Res., 47, W02531,doi:10.1029/2010WR009505.

Skøien, J.O., Blöschl, G., Laaha, G., Pebesma, E., Parajka, J., Viglione, A. (2014)Rtop: an R package for interpolation of data with a variable spatial support, with anexample from river networks, Computers & Geosciences.

Parajka, J. and Viglione, A. (2012). TUWmodel: Lumped hydrological modeldeveloped at the Vienna University of Technology for education purposes.. R packageversion 0.1-2., http://CRAN.R-project.org/package=TUWmodel.

106

Drought Signatures in the tree-ring records of EuropeanRussia

Vladimir Matskovsky1, Olga Solomina1, Sergey Matveev2, Alexander Chernokulsky3

djdfy@pochta.ru

1 Institute of Geography RAS2 Voronezh State Academy of Forestry Engineering

3 Institute of Atmospheric Physics RAS

Droughts are among the most dangerous for biosphere and agriculture naturalhazards connected with atmosphere circulation anomalies. Severe famines of1920s and 1930s in the U.S.S.R. were associated with the extended spring andsummer droughts. Droughts and heat-waves of 1970s and 2010 in the Euro-pean part of Russia result in forest fires and increase of mortality in Russianmetropolises (Lupo et al., 2012). Despite of this importance, a few importantissues connected with droughts formation are still unresolved. Particularly, it isnot clear how drought occurrence may change with the global climate changeand climate variability (Mokhov et al., 2005; Mokhov and Timazhev, 2013), andwhat is the quantitative role of the natural variability and the anthropogenicforcing in the drought formation over European Russia.

For better understanding of the drought formation mechanisms and droughtoccurrence in European Russia long records of high-resolution drought historyare necessary. Tree-ring records can be a key to resolve this issue. Differentefforts to reconstruct decadal variability of precipitation and hydrological historyof the region of the last eight centuries were previously made on the basis ofhistorical data by Swets (1978), Barash (1988), Klimenko and Sleptsov (2003).McDonald et al. (2007) used tree-ring records to reconstruct annual runoff ofthe major rivers of the northern European Russia. No tree-ring based droughtreconstructions in the East-European Plain exist so far.

Based on the analyses of 74 tree-ring chronologies we’ve shown that conifersin European Russia are sensitive to soil moisture deficit and Palmer DroughtSeverity Index to the south of 55◦N (Matskovsky and Solomina, 2011). Avail-able absolutely dated chronologies in this region are up to 300 years long, but anumber of floating chronologies from achaeological material for the last millen-nium also exist. It was shown previously that early and late wood of conifers

107

should be analyzed separately to extract information of drought season: springdrought signature can be found in earlywood while summer droughts correlatewith latewood width. The signatures of severe droughts of 1938-39, 1971-72 and2009-10 are found in tree-ring records of Voronezh region (Matveev et al., 2012).

The aim of this project is to further extend our tree-ring network to the southof European Russia and to extract spatial annually-resolved drought history fromit. The reaction of different tree species to the heat wave and drought occurredin the central and southern parts of the Russian plain in 2010 will be consideredin order to better understand the predictability of the drought using the tree ringapproach in this area. Our main goal is to reconstruct spatial drought historyfor the last three centuries based on tree-ring records.

108

A comparative study of dryness indices over Europe

Concepcion Rodriguez-Puebla1

concha@usal.es

1 Department of Atmospheric Physics, University of Salamanca, Spain

Dryness has increased over the Mediterranean area in line with a warming period.Normally, dryness is represented as a function of precipitation and evaporation.One of the main problems of dryness formulation derives from the appropriatecharacterization of evaporation. We propose simpler index to characterize dry-ness, the difference between mean monthly temperature multiplied by 2 in oCand monthly precipitation in mm (2T-P); it is derived from the interpretation ofthe climograph. Therefore, we compare results of this index with other classicalrepresentation for dryness, such as the difference between potential evaporationand precipitation (PET-P). The former approach consider only precipitationand temperature, which are variables with long historical records and facilitatecomparisons between models and observations. However, it only includes tem-perature to characterize the evaporative demand of the atmosphere. Therefore,the comparison tries to demonstrate whether the effects of humidity, radiationand wind are insignificant when analysing variability on a long time scale. Griddatasets over land of precipitation, temperature and evaporation from CRUTS3.10 are used to derive the dryness indices. The study is performed with thestandardized local departure from normal conditions. Comparisons of these twoindices allow us to infer the effect of historical warming on dryness over Europe.We approach the problem over different areas by applying Empirical OrthogonalFunction (EOF) analysis. Then, we extract modes of temporal variability andthe trend by applying the Empirical Mode Decomposition method.

109

Analysis of uncertainty in climate change projections ofdrought indices for the Czech Republic

Martin Hanel1, Adam Vizina2, Adri Buishand3

hanel@vuv.cz

1 Czech University of Life Sciences, Prague; T. G. Masaryk Water Research Institute,Prague

2 T. G. Masaryk Water Research Institute3 Royal Netherlands Meteorological Institute

Climate change scenarios for 130 catchments in the Czech Republic have beenderived with an advanced delta change method using large number of CMIP5simulations. This method transforms the observed series of precipitation andtemperature in order to result in the same changes in mean and variabilitythroughout the year as in the climate model simulation. Hydrological modelBilan was used to simulate runoff, base flow and other components of hydrologi-cal balance for the control (1960-1995) and two scenario periods (2021-2050 and2071-2100). Changes between the periods were assessed for various indices de-scribing annual and seasonal average hydrological conditions as well as drought.An extreme value model was used to characterize the distribution of deficit runoffvolumes. The changes of the parameters of this model were assessed also. Alinear mixed-effects model was used to discriminate the sources of uncertaintyin the projected hydrological changes, i.e. the uncertainty related to emissionscenario, climate model and natural variability. The results of this uncertaintyanalysis are presented and the differences with respect to the catchment charac-teristics are discussed.

110

Scaling of low flows in Czechia

Ondrej Ledvinka1

ledvinka@chmi.cz

1 Czech Hydrometeorological Institute, Prague, Czech republic

Although some national-wide studies on trends in series of hydrological droughtcharacteristics were conducted in Czechia, each of them utilized either the trendtests that do not account for the persistence at all or the modifications consid-ering short-term persistence only. In fact, similarly as in the case of short-termpersistence, also long-term persistence, often manifested through scaling, mayadversely influence the results brought by trend tests. Presence of both short-term persistence and long-term persistence means that the independence amongdata, which is required prior to the application of trend test, is violated. In par-ticular, the variance of test statistic is altered, which causes the null hypothesisof no trend is rejected too often on one hand, or rarely on the other hand.

The series of 7-day low flows together with the series of their occurrence datesat 144 water-gauging stations from all over Czechia during the hydrological pe-riod of 1961–2005 were tested for trends by Fiala et al. (2010). The investigationwas done separately for the whole year, defined specifically for these purposes,and for summer and winter seasons to distinguish between droughts triggered byevapotranspiration and droughts triggered by freezing. In addition, the authorsassessed the series comprising of deficit volumes and of numbers of days underthe threshold delineated by M-day flows Q330d and Q355d. The Mann-Kendalltest was employed and the abovementioned problem was figured out via theBayley-Hammersley-Matalas-Langbein-Lettenmayer equivalent sample size ap-proach, which solely accounts for short-term persistence. However, as stated inKhaliq et al. (2008), long-term persistence may be present in series of hydrolog-ical drought characteristics. The aim of current study was therefore to discoverwhether long-term persistence could have an influence on findings in Fiala et al.(2010).

Scaling behaviour can be summarized by means of dimensionless Hurst ex-ponent ranging from 0 to 1. Values of this exponent falling into the intervalbetween 0.5 and 1 indicate persistence. Here, due to short lengths of the series,

111

the Hurst exponent was estimated as the fractional differencing parameter inFARMA model, to which the value of 0.5 was added. To better understandwhat kind of stochastic process generates the series, so-called unit root testssuch as the Phillips-Perron test and the Kwiatkowski-Phillips-Schmidt-Shin testwere jointly applied. Persistence is distinct at more than half of the stations,no matter whether annual or seasonal series are studied. However, the shareof stations with persistence rapidly decreases in winter period. An exceptionapplies to the series of dates of occurrence, in which almost no persistence canbe recognized. The numbers of series likely generated by a long memory processare as follows: 38 (annual and summer 7-day low flows), 17 (winter 7-day lowflows), 32 (both kinds of deficit volumes), 35 (numbers of days with dischargeunder Q330d), 25 (numbers of days with discharge under Q355d), 31 (dates ofannual 7-day low flow occurrence) and 17 (dates of summer and winter 7-day lowflow occurrence). These series seem to be clustering in southwest and northeastCzechia.

112

A Statistical Analysis of the temporal-spatial changingcharacteristics of extreme precipitation in Yunnan

Province from 1958-2013

Xiao Jing Yang1, Z. X. Xu1, D. P. Zuo1, L. Liu1

yangxiaojing2007@gmail.com

1 College of Water Sciences, Beijing Normal University, No.19 Xinjiekou Wai Street,Beijing 100875, China

Since the beginning of the 21st century, frequent extreme precipitation eventsoccur under background of the global climate change. It reflected the charac-teristics of climate change that drought/flood is one of the most serious disasterimpact on human beings in the number of extreme weather events. Precipita-tion is one of the most important climatic factors that can occur the droughtand flood. However, extreme precipitation events frequently taken place in Yun-nan Province in recent years. At the same time, few studies have been done inthe field of extreme precipitation assessment in this area. So this study focuseson the characteristics of extreme precipitation in Yunnan Province. Changesin extreme precipitation for Yunnan Province, southwestern China, have beenanalyzed using data from 28 precipitation stations with daily time series duringthe period 1958-2013. In order to detect possible trends, change points, peri-ods and spatial distributions of the time series analysis were performed withMann-Kendal trend test, moving t test, Mann-Kendall abrupt change analysis,Pettit test and inverse distance weighting. Results indicated that annual totalwet-day precipitation(PRCPTOT) presented a downward tendency at the rateof -11.8mm / 10a, consecutive wet days(CWD), number of heavy precipitationdays (R10 and R20) also represented a decreasing trend, while the contributionof extreme precipitation to total precipitation presented a significant upwardtrend. Consecutive dry days (CDD), R10, R20 and PRCPTOT, CWD, whichcan reflect the characteristics of average precipitation, were not detected changepoints. CWD, very wet days (R95p), extreme wet days (R99p), simple dailyintensity index (SDII) were found separately change points in 2002, 1993, 1980and 1994. These indices reflect the changes of extreme precipitation features sen-sitively and efficiently. All extreme precipitation indices had similar oscillation

113

cycles except the CDD, R99p and SDII, which had three mainly characteristictimescales in common at 28a, 22a and 13a, the corresponding cycle respectively is18 year, 14 year, 8 years. According the meteorological geography distributions,the precipitation ranks from high to low in Yunnan province was southwest,southeast, northwest, central, northeast. In spatially, besides CDD, CWD, SDII,and PRCPTOT, the other indices change trends were consistent with overall.The maximum precipitation was not in central region, however extreme droughtsand extreme precipitation risk in this region were higher than other regions. Atthe same time, other areas spatial distribution characteristics of extreme pre-cipitation is comparatively unanimous to average annual precipitation. Overall,occurrences of extreme precipitation occurred more frequently, meanwhile, ex-treme precipitation was more concentration since 2002. More research shouldfocus encounter the extreme precipitation in Yunnan province.

114

Conceptual Framework for Satellite Data Based DroughtOutlook in Korea

Tae-Woong Kim1, Ji Yaes Shin1, Joo-Heon Lee2, Hyun-Han Kwon3

twkim72@hanyang.ac.kr

1 Hanyang University; 2 Joongbu University; 3 Chonbuk National University

Due to climate change and inequality of water availability the severity and fre-quency of drought is increasing, resulting social and economic damage in manycountries. Drought is a reoccurring and region-wide phenomenon with spatialand temporal characteristics that vary significantly from one region to another.Unlike other natural disasters, drought is hard to detect because it creeps slowlyand affect significant area. Drought is also intimidating because it already re-sults in serious damage before it is detected. With the progress of technologymany countries are recently trying to utilize satellite data to obtain reliable spa-tial and temporal information of drought. The Korean government is beginningto understand the usefulness and availability of satellite data for constructinga national drought mitigation plan. This paper presents the conceptual frame-work for satellite data based drought outlook in Korea which is currently underdevelopment. The framework contains various modules to be incorporated inthe comprehensive system for nation-wide hydro-meteorological monitoring andrequires active research works including (1) development of hydrological droughtindices based on satellite data, in which the availability of variable satellite obser-vations from MODIS, NOAA, AVHRR, etc, is investigated, (2) development ofhydrological drought monitoring system using the satellite data driven droughtindices, in which multi-satellite data are available in conjunction with broadsurface hydrological observations, (3) development of hydrological drought pre-diction model based on statistical models, in which various statistical modelssuch as artificial neural networks, hidden markov model, etc, are investigatedto verify their prediction performances, (4) development of hydrological droughtoutlook system incorporating drought monitoring system and drought predic-tion model, in which the hydrological drought index is estimated in long-term(month, season) ahead. This study also introduces the technical road-map to beperformed in nest five years.

115

Current and future agricultural droughts in thesoutheastern Mediterranean region

Lucas Menzel1, Tobias Törnros1

lucas.menzel@geog.uni-heidelberg.de

1 Department of Geography, Professorship in Hydrology and Climatology, HeidelbergUniversity

In the southeastern Mediterranean region, the annual precipitation is low andthe strong interannual climate variability recurrently triggers drought conditions.The impacts of these droughts are not only evident through sinking lake levelsand decreased river runoff, severe impacts also affect the agricultural sector. Toquantify the magnitude of these droughts, a drought index can be used. Inthis study, the performance of the Standardized Precipitation Index (SPI) andthe Standardized Precipitation Evapotranspiration Index (SPEI) was assessed.These indices can be applied on different time-scales (for example 1, 3, and 6months) in order to address different kinds of drought. The focus of this studywas on agricultural (soil moisture) drought in the wider Jordan River region, lo-cated in the south-eastern Mediterranean. Precipitation and temperature datawere spatially interpolated and used as input for the SPI and SPEI which wereapplied on multiple time-scales. Correlation analyses were thereafter conductedbetween the drought indices and the Normalized Difference Vegetation Index(NDVI) received from remote sensing. With these analyses, positive relationsrevealed that the drought indices were able to reproduce variations in vegetationgrowth. Hence, it was shown that the indices are suitable for monitoring agricul-tural drought. By using this approach, the most appropriate time-scale for thedrought indices could also be identified. In a following analysis, the indices wereapplied on data obtained from the combination of several Global- and RegionalClimate Models. By doing that, projected future drought conditions could becompared to the current conditions. The results of this study showed that thepercentage of time with moderate, severe and extreme drought conditions isprojected to strongly increase. Furthermore, a large difference could be seenbetween the SPI and SPEI when characterizing future droughts. In a final step,the irrigation water demand was simulated by applying a hydrological model on

116

a spatial scale of 1 km. The model results showed large increases of the irrigationwater demand. Hence, the results indicate that the drought vulnerability of theagricultural sector is expected to be intensified in the future.

117

Towards pan–European drought risk maps: comparing animpact-based approach with a factor-based vulnerability

assessment

Veit Blauhut1, Julia Urquijo2, Itziar Gonzalez Tanago2, Kerstin Stahl1, LuciaDe Stefano2, Mario Ballesteros2

veit.blauhut@gmail.com

1 Chair of Hydrology, University of Freiburg, Germany2 Departamento de Geodinámica, Facultad de Ciencias Geológicas Universidad Complutense

de Madrid (UCM) Madrid, Spain

Drought risk analyses require information on drought hazard together with in-formation about potential loss or damage. The magnitude of the latter is closelylinked to the specific impacts a drought has on different sectors and to the vul-nerability to drought of the assessed system. Drought impact reports gatheredin the European Drought Impacts Inventory (EDII) can serve as a proxy iden-tifier for past vulnerability of specific sectors (impact categories) to a specificdrought hazard. Based on EDII impact data for selected sectors (e.g. agricul-ture, energy and industry, water supply, water quality) the likelihood of impactoccurrence was modeled statistically for certain drought types that are typicalfor different European regions. These typical droughts were defined in terms ofthe standardized precipitation index (SPI) and estimates of drought risk in termsof impact likelihood were made and mapped. Depending on the considered sec-tor, longer drought durations of SPI were considered for southern Europe thanfor central and northern Europe. To cope with the multifaceted characteristicof the European continent, the study tests different spatial scales, starting withEuropean macro-climate regions down to smaller scale geographical and politicalunits. This work compares the risk estimated based on past drought impactswith results from a factor-based assessment of vulnerability to drought. Possiblediscrepancies between the two serve as a starting point to refine the factor-basedassessment and also to drive further refinement of the collection of past impactdata. It will thus explore how the two approaches can complement each otherto inform drought risk understanding and management across different regionsin Europe.

118

Different approaches to drought capture – how do aquaticinvertebrates indicate dry episode in Central European

streams?

Marek Polášek1, Petr Pařil1, Světlana Zahrádková1, Vít Syrovátka2, LenkaŠikulová2, Michal Straka3, Denisa Němejcová1

marek_polasek@vuv.cz

1 T. G. Masaryk Water Research Institute, p.r.i., Brno, Czech Republic2 Masaryk University Brno, Faculty of Science, Department of Botany and Zoology, Czech

Republic3 WELL Consulting, s.r.o., Brno, Czech Republic

Global climate change is often associated with changes in distribution of precipi-tation within season and changes in temperature regime which brings to CentralEurope new phenomena, like more frequent floods and drying up of watercourses.The drying up the small brooks and rivers (up to the 4th Strahler’s order) andretrospective indication of dry episode are main topics of the project "Drying upof streams during climate change" (acronym BIODROUGHT).

Basically, there are several methods how to reveal the dry episode in streamhistory. Conventional methods use different types of measuring devices, likewater level loggers and for special purposes we can use also time-lapse cameras.Other way how to indicate dry episode is to use hydrological model based onclimatic and geographical data. The idea of retrospective biological indicationis a novel approach to reveal abruption of surface flow in streams using theanalysis of taxonomic and functional composition of benthic macroinvertebrateassemblages. This method involves metrics able to quantify the frequency andextent of drying up of streams ranging from permanent to intermittent ones.

This contribution compares these different methods for indication of dryepisode in small watercourses. The dataset used for the comparison consistedof data from (i) a former state monitoring of small watercourses in 1996–2010and (ii) an ongoing research within the BIODROUGHT project (2012–2013).Thepreliminary results indicate the possibility of distinguishing intermittent and per-manent streams on the basis of presence or absence of specific macroinvertebratetaxa (bioindicators). In this study within macroinvertebrate community we fo-cused on three sensitive taxonomical groups: mayflies, stoneflies and caddisflies

119

(so called EPT taxa). Drought sensitive species are missing in the intermittentstreams and therefore they could be considered as permanency indicators. Onthe other hand, there are some taxa able to successfully survive the dry episodeand these ones occur regularly in intermittent streams. Also representation ofspecific species traits and indices calculated exclusively from EPT assemblagescan be successfully used for good distinguishing between drought impacted andpermanent sites.

This study was supported by the Technology Agency of the Czech Republic(No. TA02020395) and Project No. MUNI/A/0888/2013.

120

Atmospheric drought in Belgium - Statistical analysis ofprecipitation deficit

Orhan Dengiz1, Markéta Miháliková 1,odengiz@omu.edu.tr

1 Soil Science and Plant Nutrition Department, Agricultural Fakulty, Ondokuz MayisUniversity, Atakum, Samsun, Turkey

2 Department of Water Resources Faculty of Agrobiology, Food and Natural Resources,Czech University of Life Sciences,Prague

The field capacity and permanent wilting point are crucial parameters in hy-drological modelling. The main aim of the this research was to determine pe-dotransfer functions for estimation of basic hydraulic properties in MadendereWatershed located at Marmara region in Turkey. The study area covers about5.5 km2 and its sea level altitute is 415 m. Average annual precipitation andtemperature of the study area are 730.4 mm and 11.3 ◦C, respectively. Primaryland uses of the study area are forest, cultivated land, pasture, orchard and set-tlement. Soils of the watershed were classified as entisol, inceptisol, alfisol andultisol according to Keys to Soil Taxonomy. Two approaches were applied, thenon-linear regression and k-nearest neighbor methods in order to estimate andevaluate the hydraulic parameters . The non-linear regression performed withsatisfactory results, but the k-nearest neighbor approach performed much bet-ter. With this approach, the regression coefficients for the field capacity and thewilting point were 0.943 and 0.941, respectively. RMSD for the field capacity is1.9% vol. and for the wilting point is 2.4% vol.

121

Testing of Linear Trend Appearance in Precipitation andTemperature

Roman Kožín1,2,roman_kozin@vuv.cz

1 T. G. Masaryk Water Research Institute, p.r.i., Prague, Czech Republic2 Dept. of Water Resources and Environmental modeling, FES, CULS, Prague, Czech

Republic

The paper deals with a trend analysis that was done in the project “Review ofGroundwater Resources in the Czech Republic”. The purpose of the analysis wasto find out if there is any significant trend in precipitation and temperature andif so, when the trend probably appeared. Precipitation and temperature fromthe period 1960 – 2010 were interpolated for every of the 53 hydrogeologicalregions. To overcome seasonality the variables were aggregated into annual timestep.

For the trend analysis the standard linear model in R environment was used.The disadvantage of using such a model for this purpose is that the residuals haveto have normal distribution. That was visually tested by “Normal Q-Q plot”and all time series passed sufficiently. A method of maximum-type statistic wasused to detect a probable year of trend appearance. The method is described forinstance in Jarušková (1998) or Antoch et al. (2002). To decide if a calculatedpoint of trend appearance is significant or not one has to determine a criticalvalue of the maximum-type statistic distribution. If there is small number ofobservations the best method how to do that is to use bootstrap method. Somebootstrap methods and their robustness are described e.g. in Wilcox (2010).A testing of how many samples is needed in order to meet a balance betweenaccuracy and computational effectiveness was made as well. The distributionfunction of maximum type statistic was obtained from 20,000 randomly chosensamples with length of 50. The critical value at significance level 0.05 wascalculated from the distribution function. The critical value was consequentlyadjusted according to methodology in Jarušková (1996) because of an impact oflag (1) autocorrelation.

As a result it was found out that in precipitation there was not any significanttrend. On the other hand in case of temperature the significant trend appeared

122

in 1963 in 25 cases, 1970 just one case, 1980 in 19 cases and in 1985 in 8 cases.The slope of the trend line was 0.03 ◦C/year in 22 cases, 0.04 ◦C/year in 19 casesand 0.05 ◦C/year in 12 cases. Overall, it can be sad, that statistically significanttrend of temperature appeared on all 53 hydrological regions between years 1963and 1985 and the slope was in average 0.04 ◦C/year.

References:Antoch J., Hušková M., Jarušková D. (2001) Off-line statistical process control,

Multivariate Total Quality Control, Springer, Heidelberg, p. 1 – 86.Jarušková D. (1998) Testing appearance of linear trend, Statistical Planning and

Inference 70, p. 263 – 276.Jarušková D. (1996) Change-point detection in meteorological measurement, Monthly

Weather Review 124, p. 1535 – 1543.Wilcox R. (2010) Fundamentals of Modern Statistical Methods, Substantially Im-

proving Power and Accuracy, Second Edition

123

A project of the hydrological drought monitoring in theCzech Republic

Radek Vlnas1, Adam Beran 1,2

vlnas@vuv.cz

1 T. G. Masaryk Water Research Institute, p.r.i., Prague, Czech Republic2 Dept. of Water Resources and Environmental modeling, FES, CULS, Prague, Czech

Republic

Drought is one of the major natural hazards, which is primarily caused by defi-ciency in precipitation over an extended period.

While drought monitoring and mitigation is attracting growing attention, itis difficult to objectively quantify drought characteristics like severity, magnitudeand duration in spite of many indeces being developed for this purpose withinthe last decades. In many countries, the range of indeces is being used to quan-tify drought, often consisting a mixture of analogical indeces on various timescales and category bounds. During recent years, Standardized PrecipitationIndex (SPI) has been widely accepted and adopted in many monitoring systemsto assess the deviation from normal precipitation. The system may further in-clude percent of normal or quantile classes to evaluate drought occurrence instream flow or ground water observations. In recent years, the development ofStandardized Precipitation Evapotranspiration Index (SPEI) and StandardizedRunoff Index (SRI) can be perceived as an attempt to simplify this and use acommon basis of data processing among climatological and hydrological parts ofthe hydrological cycle.

Therefore, we suggest a drought monitoring system based on SPI method ofevaluation, which is essentially a standardizing transform of the probability ofthe observed precipitation, temperature, stream flow, borehole levels and springyields on the weekly time scale using the common clasification. The close relationbetween the SRI and the threshold level approach can be shown.

Input data transformation has been recommended as well as appropriatedistribution functions for specific type of data. Also, the web map applicationpresenting both the input data and indeces has been designed.

124

Drought propagation by using weather generators

Adam Vizina1,2, Martin Hanel1,2, Ladislav Kašpárek 1, Petr Máca1,2

vizina@vuv.cz1 T. G. Masaryk Water Research Institute, p.r.i., Prague, Czech Republic

2 Dept. of Water Resources and Environmental modeling, FES, CULS, Prague, CzechRepublic

Extreme hydrological events that have taken place at the end of the 20th andbeginning of the 21st century are represented by large scale floods or flash floodsand long lasting periods of drought. For this reason, modelling of impact ofclimate change on hydrological regime is very actual. Impacts of droughts canbe partially attenuated, however it is necessary to better understand its process.There are still lots of obstacles in drought research. Since there is no singledefinition of drought, there is currently no possibility to determine a single in-dicator or method of drought evaluation. Methods for drought evaluation arealways based on the used definitions and conditions in the locality.

This study deals with the modelling of the hydrological balance, weathergenerators, quantification and analysis of drought for current and future periods.The aim of this study is to determine the relevant meteorological index forprediction of hydrological drought in the watershed.

The first part of the study is focused on developing of weather generatorsand generating synthetic 500 years time series with 4 generators for current andfuture periods (2025, 2055 and 2085). Each generator has differ computing struc-ture. The outputs of the generators were used as inputs into three hydrologicalmodels Bilan, GR4J and TUW-HBV. On the results of the hydrological mod-elling were quantified deficit volumes and various indices of meteorological andagronomic drought. These results were used to analyse drought propagation,which is based on correlation analysis between the meteorological indices SPIand SPEI for 1, 3, 6 and 12 months with deficit volumes. For the evaluation wasalways selected 10 of the most extreme episodes in the 500 years time series foreach scenario and generator. With this combination was created dataset of 400events for each period and then were evaluated correlation analysis. Based onthe results were recommended for each basin meteorological indices.This studywas supported by the Technology Agency of the CR (No. TA01020508).

125

Drought propagation analysis using different thresholdlevel methods

Benedikt Heudorfer1, Kerstin Stahl1

benedikt.heudorfer@gmx.de

1 University of Freiburg, Freiburg, Germany

The Threshold Level Method is a useful approach to quantify droughts. It en-ables comparability across all hydrological levels and is currently being widelyused, especially in studies on drought propagation. However, the effect that thechoice of the threshold level method has on drought characteristics and droughtpropagation features, has yet been rudimentary explored. Also, most droughtpropagation studies have analysed modelled data, suggesting that applicabil-ity to observations be tested. Therefore, the aim of this study was to analysewhether, and to what extent, deviations from theory occur if different thresholdlevel method variations are applied. We tested the Constant and the VariableThreshold Level Method on precipitation, streamflow, and groundwater data,and compared drought characteristics and drought propagation patterns betweenboth methods. Thereby, also the effect of catchments controls was evaluated.

In most cases, drought propagation patterns confirmed the previously foundpropagation order from surface water to groundwater and the pooling and pro-longing of drought durations. However, we also found exceptions. While slowreacting lowland catchments with extensive aquifers performed as expected inmost cases, results from flashy, small mountainous catchments showed varia-tions with at times significant alterations of propagation patterns, depending onthe threshold level method applied. The occurrence of these alterations is at-tributed mostly to groundwater-specific catchment controls such as topographicconfinement or the relative location of a groundwater gauge within the recharge-discharge spectrum of an aquifer. In conclusion, the results suggest that in somehydrological systems, drought propagation patterns can depend strongly on thetype of threshold method and the applied threshold level. To enable suitablechoices, further studies should be conducted to determine and quantify theseeffects for a wider range of hydrological settings.

126

Atmospheric drought in Belgium - Statistical analysis ofprecipitation deficit

Sepideh Zamani1,sepideh.zamani@meteo.be

1 Royal Meteorological Institute of Belgium (RMI), Researcher, Belgium

Drought can have a substantial impact on the forest and agricultural ecosystemsof an affected region. Drought is recognized as a disaster in Belgium from re-turn periods of 20 years onwards. A more than 50 % of the land is occupiedby managed ecosystem (agriculture, forestry) in Belgium. Hence, for adequaterisk management hazard maps for drought events with a 20 year return periodare needed. The spatio-temporal characteristics of atmospheric drought are ex-plored using precipitation deficit maxima during the summer period (1 April-30September). Precipitation deficit is defined as a balance between precipitationand evapotranspiration calculated for free water surface, short grass, deciduousforest and coniferous forest.

Extreme Value Theory (EVT) [1,2] as a branch of probability and statistics,is dedicated to characterize the behaviour of extreme observations. The tailbehavior of the EVT distributions provides important features about returnlevels. EVT distributions are applicable in many study areas such as: hydrology,environmental research and meteorology, insurance and finance.

The results obtained from mapping of return levels of extreme precipitationdeficit in Belgium are compared for two different spatial modeling approaches:universal kriging and smooth GEV (generalized extreme value) using the com-bination of distance to the sea and mean summer rainfall as covariate. Thebehavior of the fitted spatial GEV-distribution is heavy-tailed with a constantshape parameter ξ 0.3 over Belgium. A comparison between both models con-firms the reliability of the smooth GEV model and shows a high similarity withboth rainfall and evapotranspiration isoline maps.

References:[1] Coles, S. (2001), "An introduction to Statistical Modelling of Extreme Values".

Springer-Verlag Heidelberg, Germany.[2] Berlant, J., Geogebeur, Y., Serger, J., Teugels, J. (2004). "Statistics of Ex-

tremes". John Wiley and Sons, Ltd.

127

The European Drought Impact report Inventory (EDII):major drought events

Irene Kohn1, J.H. Stagge2, V. Blauhut1, L. M. Tallaksen2, S. Bachmair1, K. Stahl1

irene.kohn@hydrology.uni-freiburg.de

1 University of Freiburg, Freiburg, Germany2 University of Oslo, Oslo, Norway

The range of drought impacts affecting the environment, society and economy iswide. This study presents an assessment of the diversity of drought impacts ofpast drought events across European geoclimatic regions based on data from thenew European Drought Impact report Inventory (EDII). This database, whichalready contains over 2000 entries, was developed within the EU FP7 projectDROUGHT-R&SPI. Collected impact reports are classified into 15 impact cat-egories, including different sectors, and sub-types, and referenced in time andspace to allow mapping and statistical analysis. The selection of past droughtevents studied in detail is based on the complementary European Drought Ref-erence (EDR) database, which consolidates detailed information about histor-ical large-scale drought events in Europe using climatological and hydrologicaldrought indicators (http://www.geo.uio.no/edc/droughtdb/). In terms of a widerange of impacts reported for many regions, particularly the extensive droughtsof 1947,1976, 2003, and most recently 2011-12 are outstanding. The primary im-pacts, appear to reflect timing and area of the climatic water deficits, e.g. withimpacts reported on agriculture, energy and health in 2003, which peaked in ahot summer period, versus reports on water resources impacts in long-durationdroughts. The visualization of the current EDII dataset suggests further differ-ences and commonalities in impacts between regions and events, which may berelevant for drought. Though currently somewhat biased by increasing reportavailability for more recent years and geographically by a wealth of reports fromspecific countries, the database is a promising source to assess regional and tem-poral differences in drought impacts across Europe and will continue to improveas the number and coverage of drought impacts increases with the publicallyavailable EDII website.

128

Investigating seasonal variables for droughts in coldclimates: a comparative study in Austria and Norway

Stefan Willem Ploum1, Anne Frederike van Loon2

stefan.ploum@wur.nl

1 Wageningen University, Netherlands2 University of Birmingham, United Kingdom

Droughts are often related with low flow and arid regions. This study inves-tigates hydrological droughts in cold climates during high flow season. Due tothe high amounts of water considered with peak flows, a relative small devia-tion from normal can still cause a substantial drought deficit in water supplyto downstream power plants and reservoirs. Data from 15 catchments in Aus-tria and 4 catchments in Norway has been analysed to identify the variablesthat contribute to snowmelt droughts and glaciermelt droughts. In snow influ-enced regions with mild winters, temperature and precipitation during winterare important for the development of snowmelt droughts in following spring.In cold climates, such as Norway, snowmelt droughts occur sporadically. Inglaciated catchments summer temperatures and winter precipitation are relatedto glaciermelt droughts in summer. In catchments with large glacier cover, sum-mer temperature seems more important than winter precipitation. Overall, wecannot pinpoint one variable that determines the investigated droughts. Thecomplexity of hydrological systems due to storage in snow and ice in differentclimate and geographic setting, makes it difficult to produce a general applicablerule that explains droughts in cold climates.

129

Application of trivariate Gumbel–Hougaard copula forshort term prognosis of extreme low-flow events

Wojciech Jakubowski1,Wojciech.Jakubowski@up.wroc.pl

1 Wrocław University of Environmental and Life Science, Poland

The sequences of low flows extracted by the POT method for the given thresh-old level are investigated. Each of the sequence events is described by threeindices: deficit - D, duration - T and the lowest outflow - M. For estimating thedistribution of the triple (D,T,-M) the trivariate Gumbel - Hougaard copula isselected. It is assumed that the marginal distributions are the one dimensionalGEV ones.

Estimation of copula distribution is carried out in two steps:- Unknown parameters of each of marginal distributions are estimated by max-imum likelihood method;- Parameter v of the generating function of copula is also estimated by maximumlikelihood method.

To examine the goodness-of-fit of the estimated GEV distributions as well ascopula with the observed indices the chi2 test is applied.

To describe of the low flow prediction the conditional distributions are inves-tigated. The condition is set on the level of the low flow lowest outflow. As anoutcomes it is obtained: - the expected values of conditional low flow durations(low flow deficit);- the confidence intervals for above conditional expected values;- the most probable intervals of low flow durations or deficits (on given confidencelevel).

Some results verified on independent data are calculated for the two Polishgages: Międzylesie - Nysa Kłodzka River (hilly region) and Zbytowa - WidawaRiver (Polish lowland).

130

Optimization of the Standardized Precipitation Index(SPI) for operational drought monitoring

Willem Maetens1, Brecht Ingels1, Willem Defloor1, Kris Cauwenberghs1

w.maetens@vmm.be

1 Flemish Environment Agency, Operational Water Management Division, Belgium

The Standardized Precipitation Index (SPI) is one of the most widely used in-dicators to characterize droughts. Its main advantage is that through the stan-dardization of the SPI, different locations with different climatic characteristicscan be directly compared. Hence, the SPI is also suited foroperational droughtmonitoring. The European Drought Observatory for instance uses the SPI asthe main climatological drought indicatorin a larger set of drought indicatorstomonitor drought.

However, the original definition of accumulation periods in the SPI is basedon monthly precipitation depths and allows the calculation of only one SPIvalue per month. This limits the use of the indicator for operational droughtmonitoring at a high temporal resolution. Since the original development of theSPI, large-scale and long-term daily precipitation data have become increasinglyavailable. However, the concept of a rainfall accumulation period also allows thecalculation of the SPI on a daily basis. A daily calculation of the SPI indicatorwas implemented in the framework of the Flemish pilot of the Interreg IVbproject benefit of governance for DROughtadaPtation (DROP) project. Thismore flexible definition of the accumulation period improves its usefulness asan operational drought indicator. Furthermore, it increases the potential of theSPI to better represent specific domains of interest such as soil moisture or plantwater stress through the calibration of the accumulation period to the domainof interest.

While the implementation of a daily updated SPI showed considerable po-tential for drought monitoring and as an indicator for the triggering of droughtmitigation action, it also highlights some shortcomings of the indicator that havereceived relatively little attention in the literature so far. In this respect, theeffects of the choice of reference distribution, length of reference period, sta-tionarity of the reference period data, length of the accumulation period, and

131

equal weighting of precipitation in the accumulation period are discussed andillustrated with operational examples.

In addition to daily observations, rainfall predictions from the Ensemble Pre-diction System (EPS) of the European Center for Medium range Weather Fore-casting (ECMWF) forecast modelwere used to forecast SPI values up to 10 days,increasing its usefulness for drought management.

132

A comprehensive and effective representation of thedrought situation: implementation through the web portal

www.waterinfo.be

Willem Maetens1, Pieter Cabus1, Inge De Jongh1, Willem Defloor1, KrisCauwenberghs1

w.maetens@vmm.be

1 Flemish Environment Agency, Operational Water Management Division, Belgium

Droughts have different impacts on several parts of the water system such asprecipitation, evapotranspiration, soil moisture, runoff and baseflow. Also spa-tial extent, reaction and recovery times from drought differ for different parts ofthe water system. For the development of an effective drought management, itis important to deal with the complexity of spatial and temporal variation. Asno single indicator can represent this complexity, a drought monitoring toolboxshould include multiple indicators. While efforts have been made to combineall this information in a comprehensive way (e.g. the U.S. Drought Monitor orthe European Drought Observatory), a truly comprehensive and detailed rep-resentation and interpretation of the (near) real-time drought situation is stillrare. Furthermore, a comprehensive approach to drought monitoring and man-agement should also incorporate the effective communication of this complexinformation to different stakeholders with different needs for information.

In Flanders, the recently launched web portal www.waterinfo.be aims to givea comprehensive representation of the current and forecasted state of the wa-ter system, with a focus on hydrological extremes. Therefore, information onfour main categories (floods, tides, rainfall and drought) are integrated in oneweb portal. This helps the development of an integrated approach to hydrolog-ical extremes, which is especially helpful in temperate regions such as Flanderswhere droughts are often perceived to have a smaller impact than e.g. flooding.The content and structure of the web portal is adapted to the current situation(context-driven), and the specific information needs of the end users (e.g. citi-zens, water managers, decision makers or public safety institutions such as firedepartments).

133

To monitor and represent the drought situation, a number of indicators rep-resenting the state of the water system with respect to drought such as theStandardized Precipitation Index, Precipitation Deficit, Standardized Stream-flow Index, and streamflow percentile exceedance are reported daily on the webportal. Where possible, the spatial and temporal variability of these indicatorsare reported and forecasted. Furtermore, provisions for the effective communica-tion of this information, from detailed reports to short operator interpretationsand Twitter messages, are integrated in the portal.

134

SWAT-modelling of drought impact on two catchments inFlanders

Willem Maetens1, Brecht Ingels1, Willem Defloor1, Kris Cauwenberghs1

w.maetens@vmm.be

1 Flemish Environment Agency, Operational Water Management Division, Belgium

Droughts are complex phenomena with impacts on different levels of the watersystem. Based in the impact of droughts, the effects can be subdivided in me-teorological droughts (relating to precipitation), agronomical droughts (relatingto plant and crop growth) and hydrological droughts (relating to streamflow).For each of these, different indicators and methods for the assessment of theimpact of a drought have been developed. However, it remains difficult to assessthe interactions between meteorological, agricultural and hydrological droughtsand the specific impacts of these droughts such as plant drought stress, yielddeprivation, and lack of soil moisture or streamflow in smaller streams. In theframework of the Flemish pilot of the InterregIVb project benefit of governancefor DROughtadaPtation (DROP) project, SWAT models were constructed fortwo agricultural areas in Flanders: part of the catchment of the Velpe river (94km2), and part of the catchment of the Dommel river (113 km2). These modelsare used to gain insight in the relation between drought indicators and droughtimpacts, and explore the factors that control these relations such as land use,soil type and crop type. Together, these two catchments representthe soil typesdry, moist and wet sand, dry, moist and wet sandy loam, dry, moist and wetloam, moist loam, moist heavy clay, anthropogenic soil, and peat. Modelled croptypes include winter wheat, sugar beet, potato, temporary pasture, permanentpasture, maize, orchards, mixed forest, and natural vegetation (other than for-est). Both models were calibrated using the SWAT-CUP software. In a firststage, the models were calibrated for streamflow discharge and in a second stagefor crop yield obtained from the Belgian National Insitute for Statistics (NIS).Running the models for observed historical precipitation data at the differentlocations shows the impact of droughts on crop yields, with modelled yield de-privations for the driest year in the records (1976) up to half that of the averageyield, depending on the soil and crop type. Using these models, also the impact

135

of a drought occurring at different times of the growing season, with differentlengths and different intensities can be modelled. Such approach also has a largepotential for the assessment of the likely impact of climate change scenarios onthe water system in Flanders, and the development of drought adaptation andmitigation measures.

136

Analysis of two climatological data sources to apply todescription of the precipitation behavior in an arid area:

Case of Region of Tarapacá, Northern Chile

Damian Ferney Cordoba Melgarejo1, Concepción Rodríguez Puebla2, JaznaRodríguez Segovia1

damian.cordoba@ciderh.cl1 Research and Development Center in Water Resources (CIDERH), Chile

2 Department of Physic of the Atmosphere from Univeristy of Salamanca, Spain

The Region of Tarapacá, is located in the extreme north of Chile, near at theAtacama desert, with a extension of 42.225,8 km2. The region has a strongtopographic contrast: a coastline about 0 m a.s.l to a long chain of mountains(Cordillera of Andes) about 5.000 m a.s.l. Also, the topography has a stronginfluence in the regional hydrological behavior, as long chain of mountains in-fluence in the precipitation originating the superficial flow in high altitudes fordescend in form of streams (strongs in the summer and low the rest of year)to recharge the main aquifers in the region. Therefore, the geographical, ge-omorphological and geological characteristics and its arid conditions lend theregion unique hydrological and hydrogeological features that generate evaporiteenvironments (salt flats) and deep aquifers with allochthonous recharge, so thegroundwater represents the only permanent water resource.

The recent population growth and economic development for ten years ago,has generated strong pressure on water resources due the increasing demand. Infact, northern Chile has a higher water demand than supply, boosting scenariosof competition and conflict over water uses. By this reason, is very importantstudy the future of available water resources associated with the populationgrowth and economic development, but principally, with the current climaticvariability defined by IPCC in AR5.

Define the variability, even the description the behavior of the main com-ponents of hydrological cycle (i.e. precipitation), is difficult with a poor datasource from meteorological data from stations distributed along the region. So,this study proposed the use of two data source to study the characterizationof the precipitation as important component of the regional hydrological cy-cle: The Climate Forecast System Reanalysis (CFSR) from National Center for

137

Environmental Prediction (NCEP) of the National Oceanic and AtmosphericAdministration (NOAA); and the Monthly Climatic Observations CRU TS3.10Dataset from Climatic Research Unit of the University of East Anglia, UK.CFSR dataset has a great number of variables, with a good temporal (hourly,dayly, monthly) and spatial (from 2,5◦ to 0,3◦) resolution, originate from cli-matological model and observations from 1976 to present. CRU TS3.10 has areduced number of variables, with a monthly resolution and 0,5◦ of spatial res-olution, originate from interpolation process of stations from Monthly ClimaticData for the World (MCDW) project, CLIMAT and the Australian Bureau ofMeteorology with data from 1901 to present.

For the study, just be define the use of data from CFSR with monthly resolu-tion and 0,5◦ of spatial resolution, to compare with CRU TS3.10 data applyingClimate Data Operators (CDO) and the tools of GrADS software. For the twodata sources, the information about the precipitation always be generated insurface level.

The result of the comparison were evaluated with four dataset from regionalmeteorological stations. The comparison between these two data source, showsthat the reanalysis data is more frgaile than the interpolation data, but is pos-sible, according with Harris et al,2013, in this region there is a major deviationsdue to sparser observational data.

138

Evaluation of the socio-economic impacts of droughtevents: evidence from some areas in Europe

Dario Musolino1

dario.musolino@unibocconi.it

1 CERTeT-Bocconi, Italy

This contribution deals with the question of the quantification of the socio-economic effects of drought events. Its objective is to evaluate the impacts ofdrought on the most relevant and vulnerable economic sectors (like agricultureand energy) in monetary terms, in four case study areas (Portugal, Po basin inItaly, Jucar basin in Spain, Syros Island in Greece). Its theoretical and method-ological background is based on the consumer surplus theory, where apparentlyprices and production are assumed to be the key economic variables.

The first part is dedicated to the presentation of the theoretical and method-ological approach. The second part is devoted to the characterization of the casestudy areas (and the related drought events that they have suffered), and to theillustration of the most relevant impacts occurred in each area. The fourth sec-tion, the most important one, presents and discusses the results of the analysesabout the economic impacts.

One of the most remarkable outcomes of such analyses is the fact that theeconomic effects of the drought events are socially differentiated: consumers andproducers, for example, were affected differently by the drought events. Thefirst group can even earn from the drought, because of the price effect causedby the scarcity of the goods produced; the second group always loses, becauseboth of the quantity effect and of the price effect. Different impacts were alsoestimated among producers, in particular when they are distinguished makingreference to the kind of products (crop, for agriculture), and to the geographicalarea. Policy implications of such results are finally discussed and presented. Thecontents of this contribution derives from the work which has been done for theVII Framework Programme project DROUGHT-R&SPI “Fostering EuropeanDrought Research and Science-Policy Interfacing“.

139

An application of remote sensing to support to waterresources management

Francesco Vuolo1

francesco.vuolo@boku.ac.at

1 University of Natural Resources and Life Sciences, Austria

Irrigated agriculture is the main user of freshwater resources (30% in CentralEurope, 60% in the South). Efficient water management is therefore of essen-tial importance, especially where water scarcity and water quality are becomingsevere challenges. To achieve a successful and effective use of resources, farm-ers and water managers require easy-to-use decision support tools and reliableinformation.

In this poster, we present the results of the EO4Water-2 project: an applica-tion in Lower Austria of remote sensing technologies to support farmers (at fieldscale) and decision makers (at catchment scale). The project aims at combiningoptical high resolution satellite data, in-situ agro-meteorological measurementsand a soil water balance model to support irrigation management. It extends theapproaches developed in the first project by addressing irrigation managementunder water stress conditions. To enhance the utilization of the limited waterresources, our tools take into account the actual crop development and the soilwater conditions to provide timely and personalised information that supportsfarmers in the irrigation process.

140

Flow-duration-frequency and duration-flow-frequencyapproaches in low-flow analysis, advantages and

constraints

Ewa Bogdanowicz1, Witold Gustaw Strupczewski1, Sisay Debele1, KrzysztofKochanek1

ewabgd@o2.pl

1 Institute of Geophysics, Polish Academy of Sciences, 01-452 Warsaw, CHIHE project

Demands for low-flow information exceed the scope of standard characteristicsprovided by Polish National Hydrological Service in the form of the minimumand the mean minimum stage and discharge. To meet these demands, two com-plementary methods for estimating low-flow frequency characteristics are pro-posed. Both methods deal with duration and flow. The first one “flow-duration-frequency” (known as QdF) has already been applied successfully to low-flowanalysis, flood flows and rainfall intensity. The second one called “duration-flow-frequency” (DqF) was proposed by Strupczewski et al. to flood frequencyanalysis and extended to the non-stationary case. The two methods differ in thetreatment of flow and duration. In the first method the duration (d-consecutivedays) is a chosen fixed value and the frequency analysis concerns the annual orseasonal series of minimum flow non-exceeded within d-day period. In the secondmethod, the flow is a fixed threshold and the duration of flows non-exceedingthis threshold is a subject of frequency analysis. The statistical properties of thetwo methods are presented with the focus on their advantages and constrains inpractical applications: to evaluation of waste dilution potential, for engineeringdesign purposes and in the management of water-quality and water-supply prob-lems. The description of present low-flow regime at selected gauged sites andthe application of two methods for assessment of potential alteration of droughtregime induced by future climate are shown.

141

Comparison of SOM and non-hierarchic clusteringtechniques by pooling of low flows in Slovakia

Kamila Hlavcova1, Silvia Kohnová1, Miroslav Sabo1, Andrea Stevková1, JanSzolgay1

kamila.hlavcova@stuba.sk

1 Department of Land and Water Resources Management, Faculty of Civil Engineering,Slovak University of Technology Bratislava, Radlinskeho 11, 813 68 Bratislava, Slovak

Republic

The aim of this study was to derive pooling groups for undirect estimationof design low flows on ungauged catchments in Slovakia. For the deriving ofpooling groups the Self-organizing maps (SOM) were used. This methodologywas compared with results of non-hierarchic K-means clustering method. Forthe analysis 217 small and mid-sized catchments from the whole territory ofSlovakia were selected. Ranging area was from 10 to 500 km2 with the period ofobservations longer than 20 years. The pooling procedure was based on groupingthe relative frequencies of occurrence of annual minimum discharges lower thenQ95 in separate months during the year, estimated in all observation stations.

The quality of the individual clustering methods (SOM and K-means) wastested using validity measures as the Dunn index, the Connectivity index, theAverage Silhouette plot (Euclidean, Squared Euclidean and Manhattan distance)and the Davies-Bouldin Index.

All of these tests confirmed grouping of the catchments into two clusters.Consequently the intersection between the SOM and the K-means clusteringmethods, using the Rand coefficient was performed. The Rand coefficient was0.991, what proved that these two clustering methods were similar, and theydiffer only in pooling of two catchments.

The statement of some recently published studies that the SOM is very sim-ilar to the K-means clustering method was also confirmed in our study. Finallywe can state that Self-organizing maps are a suitable tool for pooling of low flowsin Slovakia.

142

The inadequacy of the standardized precipitation index asa groundwater drought indicator

Jude Lubega Musuuza1, Rohini Kumar1, Anne F. Van Loon2, RolandBarthel3, Jurriaan ten Broek2, Luis E. Samaniego1, Adriaan J. Teuling2,

Sabine Attinger1

jude.musuuza@ufz.de

1 Helmholtz Centre for Environmental Research - UFZ, Leipzig2 Wageningen University

3 University of Gothenburg

Droughts occur when precipitation and storage in surface-water bodies, soil andthe ground fall below statistical average. Surface water storage reacts veryquickly to precipitation events and is prone to short-term droughts. Soil mois-ture and groundwater storage react slowly and are affected by persistent deficitsin precipitation. The development of groundwater droughts lags behind those inprecipitation and surface water. Long-term measurements of precipitation areglobally available at reasonable resolutions while soil moisture measurements areonly available for a few regions for the top few centimeters of the soil and theinvolved costs are prohibitive. Groundwater measurements are also scarce andobtaining reliable, long-term head observations is a challenge, mainly due toinfrastructural costs and anthropogenic influences. Therefore, the StandardizedPrecipitation Index (SPI) is easy and inexpensive to estimate, thus the appealingindex of choice for drought studies. We show here that in contrast, the SPI is byitself inadequate for groundwater drought studies. We present a StandardizedGroundwater Index (SGI), which is computed in a similar way as the SPI butusing groundwater head observations from two geologically very different areasin Germany and The Netherlands. We further show the pitfalls of using theSPI as a groundwater drought index. The spatial variation in catchment char-acteristics, like soil, elevation and aquifer properties results in spatial variationin groundwater drought that is not captured by SPI.

143